4D camera tracking and optical stabilization

Abstract

A light-field video stream may be processed to modify the camera pathway from which the light-field video stream is projected. A plurality of target pixels may be selected, in a plurality of key frames of the light-field video stream. The target pixels may be used to generate a camera pathway indicative of motion of the camera during generation of the light-field video stream. The camera pathway may be adjusted to generate an adjusted camera pathway. This may be done, for example, to carry out image stabilization. The light-field video stream may be projected to a viewpoint defined by the adjusted camera pathway to generate a projected video stream with the image stabilization.

Description

CROSS-REFERENCE TO RELATED APPLICATIONS

The present application is related to U.S. patent application Ser. No. 13/688,026, for “Extended Depth of Field and Variable Center of Perspective in Light-Field Processing”, filed Nov. 28, 2012, and issued on Aug. 19, 2014 as U.S. Pat. No. 8,811,769, the disclosure of which is incorporated herein by reference in its entirety.

The present application is related to U.S. patent application Ser. No. 13/774,971, for “Compensating for Variation in Microlens Position During Light-Field Image Processing”, filed Feb. 22, 2013, and issued on Sep. 9, 2014 as U.S. Pat. No. 8,831,377, the disclosure of which is incorporated herein by reference in its entirety.

TECHNICAL FIELD

The present disclosure relates to digital imaging. More precisely, the present disclosure relates to use of light-field data to track the motion path of a camera and/or adjust the stability of image data generated by the camera.

BACKGROUND

In conventional 2D digital photography, an image of a scene may be captured as a 2D matrix of color values that represents the scene from one field of view. The focus depth and Center of Perspective (CoP) of the image typically cannot be changed after the image has been captured; rather, the focus depth and Center of Perspective at the time of image capture determine what features are in focus and in view. Accordingly, there is also no way to modify the viewpoint from which an image is taken.

One repercussion of this limitation is that it may be difficult to carry out image stabilization. Since true image stabilization would require adjustment of the viewpoint from which the image or video was captured, conventional 2D image stabilization methods are typically limited to lossy processes that can only compensate for 2D shifts within an image sequence. The need to crop portions of the image to correct the 2D shifts results in loss of image data.

SUMMARY

According to various embodiments, a light-field video stream may be processed to obtain a camera pathway indicative of the viewpoint from which a light-field video stream was generated (i.e., captured). The camera pathway may be modified to obtain an adjusted camera pathway, which may provide a more desirable viewpoint. For example, the adjusted camera pathway may be stabilized relative to the camera pathway to provide image stabilization. In the alternative, the adjusted camera pathway may be de-stabilized, or “littered,” relative to the camera pathway to simulate vibration or other motion of the viewer's viewpoint.

The camera pathway may be obtained in various ways. According to one embodiment, a plurality of target pixels may be selected, in a plurality of key frames of the light-field video stream. The target pixels may have predetermined color and/or intensity characteristics that facilitate tracking of the target pixels between frames. For example, the target pixels may be selected from static, textured objects that appear in the key frames. The target pixels may further be from planar regions of the objects to further facilitate tracking.

According to some embodiments, the target pixels may be identified by generating a list of a plurality of targets appearing in each of the key frames, generating a plane model for each of the targets for each of the key frames, and then generating a mask for each of the targets for each of the key frames, indicating one or more target pixels within each of the targets. Further, superpixel segmentation may be carried out, and a motion error map may be calculated, for each of the key frames. The superpixels and motion error maps may be used to access texture and motion error for each of the superpixels for each key frame, to identify a plurality of the superpixels as candidate targets. A plane may be fitted to each of the candidate targets for each key frame. The targets may then be selected from among the candidate targets.

If desired, identification of the target pixels may be facilitated by using a depth map for each of the key frames, and/or initial camera motion, generated by a sensor operating contemporaneously with capture of the light-field video stream. The camera motion may be indicative of motion of the light-field camera during at least a segment, containing the key frames, of the light-field video stream. In some examples the camera motion may be for an initial segment of the light-field video stream, and may facilitate accurate identification and/or location of the targets.

The target pixels may be used to generate a camera pathway indicative of motion of the camera during generation of the light-field video stream. The camera pathway may have six degrees of freedom, and may encompass the entirety of the video stream. A 3D mapping of the target pixels may also be generated.

In some embodiments, the camera pathway may be generated by dividing the light-field video stream into a plurality of sequences, each of which begins with one of the key frames. For each segment, starting with the first segment, the position and/or orientation of the target pixels may be tracked in each frame, and changes in the positions and/or orientations may be compared between frames to obtain a portion of the camera pathway for that segment. The position and/or orientation of each of the target pixels in the last frame of a sequence may be used for the starting key frame of the next sequence.

If desired, generation of the camera pathway may be facilitated by using camera-intrinsic parameters obtained from calibration of the light-field camera, light-field optics parameters pertinent to one or more light-field optical elements of the light-field camera, and/or camera motion, generated by a sensor operating contemporaneously with capture of the light-field video stream. The camera motion may be indicative of motion of the light-field camera during at least a segment of the light-field video stream.

The camera pathway may be adjusted to generate an adjusted camera pathway. This may be done, for example, to carry out image stabilization. Image stabilization may be improved by adjusting U,V coordinates within each of a plurality of frames of the light-field video stream to cause frame-to-frame motion to be relatively smooth and contiguous

The light-field video stream may be projected to a viewpoint defined by the adjusted camera pathway to generate a projected video stream with the image stabilization. The projected video stream may be outputted to an output device, such as a display screen.

These concepts will be described in greater detail below.

BRIEF DESCRIPTION OF THE DRAWINGS

The accompanying drawings illustrate several embodiments. Together with the description, they serve to explain the principles of the embodiments. One skilled in the art will recognize that the particular embodiments illustrated in the drawings are merely exemplary, and are not intended to limit scope.

FIG. 1 depicts a portion of a light-field image.

FIG. 2 depicts an example of an architecture for implementing the methods of the present disclosure in a light-field capture device, according to one embodiment.

FIG. 3 depicts an example of an architecture for implementing the methods of the present disclosure in a post-processing system communicatively coupled to a light-field capture device, according to one embodiment.

FIG. 4 depicts an example of an architecture for a light-field camera for implementing the methods of the present disclosure according to one embodiment.

FIG. 5 is a flow diagram depicting a method for carrying out image stabilization, according to one embodiment.

FIG. 6 is a flow diagram depicting the step of selecting the target pixels from the method of FIG. 5, in greater detail, according to one embodiment.

FIG. 7 is a screenshot depicting superpixel segmentation of a frame, according to one embodiment.

FIG. 8 is a screenshot depicting a motion error map for the frame of FIG. 7, according to one embodiment.

FIG. 9 is a screenshot depicting identification of candidate targets with texture and small motion error, in the frame of FIG. 7, according to one embodiment.

FIG. 10 is a screenshot depicting selection of targets in the frame of FIG. 7, according to one embodiment.

FIG. 11 is a flow diagram depicting the step of generating the camera pathway from the method of FIG. 5, according to one embodiment.

FIG. 12 is an illustration depicting the relationship between camera acceleration curves and light-field optical constraints, according to one embodiment.

FIG. 13 is a screenshot depicting the use of LiDAR and gyroscope data in combination with a light-field video stream to generate more accurate camera pathways.

DEFINITIONS

For purposes of the description provided herein, the following definitions are used:

• • Adjusted camera pathway: a camera pathway that has been deliberately modified.
  • Camera pathway: a pathway indicative of motion of a camera
  • Conventional image: an image in which the pixel values are not, collectively or individually, indicative of the angle of incidence at which light is received by a camera.
  • Depth: a representation of distance between an object and/or corresponding image sample and a camera or camera element, such as the microlens array of a plenoptic light-field camera.
  • Disk: a region in a light-field image that is illuminated by light passing through a single microlens; may be circular or any other suitable shape.
  • Image: a two-dimensional array of pixel values, or pixels, each specifying a color.
  • Input device: any device that receives input from a user.
  • Light-field camera: any camera capable of capturing light-field images.
  • Light-field data: data indicative of the angle of incidence at which light is received by a camera.
  • Light-field image: an image that contains a representation of light-field data captured at the sensor.
  • Light-field video stream: a sequential arrangement of light-field data captured over a length of time, from which a video stream can be projected.
  • Main lens: a lens or set of lenses that directs light from a scene toward an image sensor.
  • Mask: a map representing whether a set of pixels possesses one or more attributes, such as the attributes needed to operate as a target pixel.
  • Microlens: a small lens, typically one in an array of similar microlenses.
  • Microlens array: an array of microlenses arranged in a predetermined pattern.
  • Output device: any device that provides output to a user.
  • Scene: a collection of one or more objects to be imaged and/or modeled.
  • Image sensor: a light detector in a camera capable of generating electrical signals based on light received by the sensor.
  • Subaperture view: an image generated from light-field data from the same location on each microlens of a microlens array or each camera image of a tiled camera array.
  • Superpixel segmentation: division of an image into groups (superpixels) of adjacent pixels.
  • Target: a portion of a light-field image, containing multiple pixels, including at least one target pixel.
  • Target pixel: a pixel of a target, with properties suitable for automated identification and/or modeling in 3D space.

In addition to the foregoing, additional terms will be set forth and defined in the description below. Terms not explicitly defined are to be interpreted, primarily, in a manner consistently with their usage and context herein, and, secondarily, in a manner consistent with their use in the art.

For ease of nomenclature, the term “camera” is used herein to refer to an image capture device or other data acquisition device. Such a data acquisition device can be any device or system for acquiring, recording, measuring, estimating, determining and/or computing data representative of a scene, including but not limited to two-dimensional image data, three-dimensional image data, and/or light-field data. Such a data acquisition device may include optics, sensors, and image processing electronics for acquiring data representative of a scene, using techniques that are well known in the art. One skilled in the art will recognize that many types of data acquisition devices can be used in connection with the present disclosure, and that the disclosure is not limited to cameras. Thus, the use of the term “camera” herein is intended to be illustrative and exemplary, but should not be considered to limit the scope of the disclosure. Specifically, any use of such term herein should be considered to refer to any suitable device for acquiring image data.

In the following description, several techniques and methods for processing light-field images are described. One skilled in the art will recognize that these various techniques and methods can be performed singly and/or in any suitable combination with one another. Further, many of the configurations and techniques described herein are applicable to conventional imaging as well as light-field imaging. Thus, although the following description focuses on light-field imaging, all of the following systems and methods may additionally or alternatively be used in connection with conventional digital imaging systems. In some cases, the needed modification is as simple as removing the microlens array from the configuration described for light-field imaging to convert the example into a configuration for conventional image capture.

Architecture

In at least one embodiment, the system and method described herein can be implemented in connection with light-field images captured by light-field capture devices including but not limited to those described in Ng et al., Light-field photography with a hand-held plenoptic capture device, Technical Report CSTR 2005-02, Stanford Computer Science. Further, any known depth sensing technology may be used.

Referring now to FIG. 2, there is shown a block diagram depicting an architecture for implementing the method of the present disclosure in a light-field capture device such as a camera 200. Referring now also to FIG. 3, there is shown a block diagram depicting an architecture for implementing the method of the present disclosure in a post-processing system 300 communicatively coupled to a light-field capture device such as a camera 200, according to one embodiment. One skilled in the art will recognize that the particular configurations shown in FIGS. 2 and 3 are merely exemplary, and that other architectures are possible for camera 200 and post-processing system 300. One skilled in the art will further recognize that several of the components shown in the configurations of FIGS. 2 and 3 are optional, and may be omitted or reconfigured.

In at least one embodiment, camera 200 may be a light-field camera that includes light-field image data acquisition device 209 having optics 201, image sensor 203 (including a plurality of individual sensors for capturing pixels), and microlens array 202. Optics 201 may include, for example, aperture 212 for allowing a selectable amount of light into camera 200, and main lens 213 for focusing light toward microlens array 202. In at least one embodiment, microlens array 202 may be disposed and/or incorporated in the optical path of camera 200 (between main lens 213 and image sensor 203) so as to facilitate acquisition, capture, sampling of, recording, and/or obtaining light-field image data via image sensor 203. The microlens array 203 may be positioned on or near a focal plane 204 of the main lens 213.

Referring now also to FIG. 4, there is shown an example of an architecture for a light-field camera, or camera 200, for implementing the method of the present disclosure according to one embodiment. FIG. 4 is not shown to scale. FIG. 4 shows, in conceptual form, the relationship between aperture 212, main lens 213, microlens array 202, and image sensor 203, as such components interact to capture light-field data for one or more objects, represented by an object 401, which may be part of a scene 402.

In at least one embodiment, camera 200 may also include a user interface 205 for allowing a user to provide input for controlling the operation of camera 200 for capturing, acquiring, storing, and/or processing image data. The user interface 205 may receive user input from the user via an input device 206, which may include any one or more user input mechanisms known in the art. For example, the input device 206 may include one or more buttons, switches, touch screens, gesture interpretation devices, pointing devices, and/or the like.

Similarly, in at least one embodiment, post-processing system 300 may include a user interface 305 that allows the user to provide input to control parameters for post-processing, and/or for other functions.

In at least one embodiment, camera 200 may also include control circuitry 210 for facilitating acquisition, sampling, recording, and/or obtaining light-field image data. The control circuitry 210 may, in particular, be used to switch image capture configurations such as the zoom level, resolution level, focus, and/or aperture size in response to receipt of the corresponding user input. For example, control circuitry 210 may manage and/or control (automatically or in response to user input) the acquisition timing, rate of acquisition, sampling, capturing, recording, and/or obtaining of light-field image data.

In at least one embodiment, camera 200 may include memory 211 for storing image data, such as output by image sensor 203. Such memory 211 can include external and/or internal memory. In at least one embodiment, memory 211 can be provided at a separate device and/or location from camera 200.

In at least one embodiment, captured image data is provided to post-processing circuitry 204. The post-processing circuitry 204 may be disposed in or integrated into light-field image data acquisition device 209, as shown in FIG. 2, or it may be in a separate component external to light-field image data acquisition device 209, as shown in FIG. 3. Such separate component may be local or remote with respect to light-field image data acquisition device 209. Any suitable wired or wireless protocol may be used for transmitting image data 321 to circuitry 204; for example, the camera 200 can transmit image data 321 and/or other data via the Internet, a cellular data network, a Wi-Fi network, a Bluetooth communication protocol, and/or any other suitable means.

Such a separate component may include any of a wide variety of computing devices, including but not limited to computers, smartphones, tablets, cameras, and/or any other device that processes digital information. Such a separate component may include additional features such as a user input 315 and/or a display screen 316. If desired, light-field image data may be displayed for the user on the display screen 316.

Overview

Light-field images often include a plurality of projections (which may be circular or of other shapes) of aperture 212 of camera 200, each projection taken from a different vantage point on the camera's focal plane. The light-field image may be captured on image sensor 203. The interposition of microlens array 202 between main lens 213 and image sensor 203 causes images of aperture 212 to be formed on image sensor 203, each microlens in microlens array 202 projecting a small image of main-lens aperture 212 onto image sensor 203. These aperture-shaped projections are referred to herein as disks, although they need not be circular in shape. The term “disk” is not intended to be limited to a circular region, but can refer to a region of any shape.

Light-field images include four dimensions of information describing light rays impinging on the focal plane of camera 200 (or other capture device). Two spatial dimensions (herein referred to as x and y) are represented by the disks themselves. For example, the spatial resolution of a light-field image with 120,000 disks, arranged in a Cartesian pattern 400 wide and 300 high, is 400×300. Two angular dimensions (herein referred to as u and v) are represented as the pixels within an individual disk. For example, the angular resolution of a light-field image with 100 pixels within each disk, arranged as a 10×10 Cartesian pattern, is 10×10. This light-field image has a 4-D (x,y,u,v) resolution of (400,300,10,10). Referring now to FIG. 1, there is shown an example of a 2-disk by 2-disk portion of such a light-field image, including depictions of disks 102 and individual pixels 101; for illustrative purposes, each disk 102 is ten pixels 101 across.

In at least one embodiment, the 4-D light-field representation may be reduced to a 2-D image through a process of projection and reconstruction. As described in more detail in related U.S. Utility application Ser. No. 13/774,971 for “Compensating for Variation in Microlens Position During Light-Field Image Processing,” filed Feb. 22, 2013, the disclosure of which is incorporated herein by reference in its entirety, a virtual surface of projection may be introduced, and the intersections of representative rays with the virtual surface can be computed. The color of each representative ray may be taken to be equal to the color of its corresponding pixel.

Camera Pathway Generation and Adjustment

There are many instances in which it is desirable to obtain the 3D pathway followed by a camera to capture a scene. For example, in order to integrate computer-generated objects or effects in a scene, it may be desirable to render the computer-generated elements with a virtual camera that remains aligned with the actual camera used to capture the scene. Further, integration of the scene with audio effects may be done with reference to the camera pathway. For example, the volume and/or speaker position of audio effects may be determined based on the camera position and/or orientation in any given frame.

It may be most helpful to obtain a camera pathway with six degrees of freedom (for example, three to specify camera position along each of three orthogonal axes, and three to specify the orientation of the camera about each axis) for each frame. In this application, “camera pathway” includes the position and/or orientation of the camera.

In addition to the uses mentioned above, obtaining the camera pathway may enable the camera pathway to be adjusted for various purposes. Light-field image capture provides the unique ability to reproject images at different Centers of Perspective, allowing the viewpoint of the camera to effectively be shifted. Further details regarding projection of light-field data may be found in U.S. Utility application Ser. No. 13/688,026, for “Extended Depth of Field and Variable Center of Perspective in Light-Field Processing”, filed Nov. 28, 2012, the disclosure of which is incorporated herein by reference in its entirety.

In some embodiments, the camera pathway may be adjusted to smooth out the camera pathway, thereby effectively stabilizing the camera. Such stabilization may not have the losses and limitations inherent in known image stabilization algorithms used for conventional 2D images. As another alternative, jitter may be added to the camera pathway, causing the reprojected view to shake. This may be used to simulate an explosion, impact, earthquake, or the like, after image capture.

FIG. 5 is a flow diagram depicting a method 500 for generating and adjusting a camera pathway to carry out image stabilization or other adjustments, according to one embodiment. The method 500 may be used in conjunction with light-field data captured by one or more plenoptic light-field cameras such as the light-field camera 200 of FIG. 2. Additionally or alternatively, the light-field data may be captured through the use of a different camera system, such as a tiled camera array that captures light-field data without the use of a microlens array.

The method 500 may start 510 with a step 520 in which the light-field video stream is captured. This may be done by a light-field camera such as the light-field camera 200 of FIG. 2, or by a different type of light-field image capture system, as mentioned previously. The light-field video stream may be the image data 321 referenced in FIG. 3.

In a step 530, the light-field video stream may be received, for example, at a processor capable of processing the light-field video stream. The processor may be the post-processing circuitry 204 of the camera 200, as in FIG. 2, and/or the post-processing circuitry 204 of the post-processing system 300, as in FIG. 3. In the alternative, any processor capable of processing light-field data may receive the light-field video stream.

In a step 540, target pixels may be selected in key frames of the light-field video stream. Target pixels may be pixels with color/intensity characteristics that make them easy to automatically recognize, and hence track from one frame to another. The target pixels may be identified, at least, in key frames of the light-field video stream. The step 540 will be described in greater detail in connection with FIG. 6.

In a step 550, a camera pathway may be generated, indicative of motion of the light-field camera used to generate (i.e., capture) the light-field video stream. If some information about the camera pathway is already available at the commencement of the step 550, the step 550 may include gathering the remaining data needed to generate the camera pathway with six degrees of freedom, for the entire length of the light-field video stream. The step 550 will be described in greater detail in connection with FIG. 11.

In a step 560, an adjusted camera pathway may be generated, based on the camera pathway. The adjusted camera pathway may include any desired adjustments, such as camera stabilization or camera jittering. This step is optional; as mentioned previously, the camera pathway may be useful independently of the creation of an adjusted camera pathway. For example, integration of computer-generated elements in the light-field video stream may not require the adjustment of the camera pathway, but may rather be based on the un-adjusted camera pathway.

In a step 570, a video stream may be projected based on the adjusted camera pathway. The video stream may be projected from the viewpoint of the camera, in each frame, as indicated on the adjusted camera pathway. The adjusted camera pathway may also provide the position and orientation of the camera with six degrees of freedom, and may thus provide the information needed to generate new projected views. The video stream generated in the step 570 may thus reflect the modifications made to the camera pathway, such as image stabilization. This step is optional, and may be unnecessary if the step 560 is not carried out.

In a step 580, the video stream generated in the step 570 may be output to an output device. This may be, for example, the display screen 316 of the post-processing system 300 of FIG. 3. Additionally or alternatively, the video stream may be output to any other suitable output device, such as a monitor or other display screen that is not part of a light-field data processing system. This step is also optional, and may not be needed if the step 560 and the step 570 are not performed. The method 500 may then end 590.

Various steps of the method 500 of FIG. 5 may be re-ordered, iterated, and/or altered in various ways. Further, various steps of the method 500 may be omitted, replaced with alternative steps, or supplemented with additional steps not specifically shown and described herein. Such modifications would be understood by a person of skill in the art, with the aid of the present disclosure.

Target and Target Pixel Identification

FIG. 6 is a flow diagram depicting the step 540 of selecting the target pixels from the method of FIG. 5, in greater detail, according to one embodiment. The step 540 will be described with reference to FIGS. 7 through 10. The step 540 may utilize one or more of the following, which may be included in the light-field video stream and/or provided separately:

• • Designation of at least two key frames (for example, a first key frame and a second key frame) in the light-field video. This designation may be made by a user, or automatically by the system. In some embodiments, the key frames may be arbitrarily selected.
  • Depth maps for each of the key frames. The depth maps may be obtained by processing the light-field video stream and/or from one or more depth sensors, such as LiDAR or time-of-flight sensors, that captured depth data synchronously with capture of the light-field video stream.
  • Initial camera motion for a sequence of frames that contains the first and second key frames. The initial camera motion need not apply to the entire light-field video stream, but may rather be applicable to only a portion, such as the initial frames of the light-field video stream. In some embodiments, the initial camera motion may be obtained from data captured by other sensors, such as LiDAR sensors, gyroscopes, accelerometers, or other sensors that measure depth, position, orientation, velocity, and/or acceleration. Synchronous location and mapping (SLAM) techniques or the like may be applied to such sensor data to obtain the initial camera motion.

The step 540 may be designed to provide output, which may include one or more of the following:

• • A list of targets, each of which is defined in at least the first frame by a closed contour. The closed contour may be a list of targets, for example, designating each target by (x, y) coordinates.
  • A plane model of each target, for example, providing the position and orientation of a plane passing through the target. The plane model may designate the plane, for example, by a normal vector n and an offset d.
  • A mask for each of the targets for each key frame, indicating one or more target pixels within each of the targets. The target pixels may be the pixels within each target that are suitable for matching in different frames.

The step 540 may utilize direct image mapping to determine the camera pose and motion, and the depth of objects in the scene. The targets used for direct image mapping may be selected to facilitate identification and matching between frames. Thus, each of the targets may have color and/or intensity characteristics that facilitate identification. The targets may advantageously be static, so that relative motion of the targets between frames can be used to ascertain motion of the camera (as opposed to motion of the targets). Further, the targets may have textures that make them relatively easy to identify with accuracy.

Further, in at least one embodiment, only planar regions (i.e., planar surfaces of objects) may be selected as targets. This may facilitate usage of planes to approximate the targets, and may minimize the number of unknowns in the expressions used to solve for depth. Specifically, for a planar region, only four unknowns need to be solved for.

As shown, the step 540 may begin 610 with a step 620 in which superpixel segmentation of each key frame is carried out. Superpixel segmentation may entail division of each key frame into groups (superpixels) in which pixels have some traits in common, such as color and/or intensity values.

FIG. 7 is a screenshot depicting superpixel segmentation of a frame 700, according to one embodiment. The frame 700 may be divided into superpixels 710, as shown. The superpixels 710 may be of a generally, but not precisely, uniform size and shape. Any of a variety of superpixel segmentation algorithms known in the art may be used. In some embodiments, superpixel segmentation may be carried out via Simple Linear Iterative Clustering (SLIC) or a similar method.

Returning to FIG. 6, in a step 630, a motion error map may be calculated for each of the key frames. The motion error map may be a grayscale representation of relative motion between frames (for example consecutive frames). The motion error map may reveal which elements of the scene are moving between the frames, and which are stationary.

FIG. 8 is a screenshot depicting a motion error map 800 for the frame 700 of FIG. 7, according to one embodiment. As shown, the motion error map 800 indicates that the people 810 in the foreground are moving, while the background elements 820 are stationary.

Returning to FIG. 6, in a step 640, candidate targets may be identified from among the superpixels 710 of the frame 700. This may be done by using the superpixels delineated in the step 620 and the motion error maps generated in the step 630. Candidate targets may be superpixels with small motion error (static) and strong gradient (texture). In some embodiments, only superpixels 710 with easily-recognizable textures, in which little or no motion between frames has occurred, may be designated as candidate targets.

FIG. 9 is a screenshot depicting identification of candidate targets 910 with texture and small motion error, in the frame 700 of FIG. 7, according to one embodiment. As shown, some of the superpixels 710 of FIG. 7 have been identified as candidate targets 910. Notably, superpixels 710 lacking in texture (such as those of the blank wall behind the people 810) have not been selected, and moving elements (such as the people 810) also have not been selected. Rather, the candidate targets 910 are portions that are generally stationary and are textured enough to be readily recognized.

Returning to FIG. 6, in a step 650, planes may be fitted to the candidate targets 910. Thus, each of the candidate targets 910 may be approximated or modeled as a portion of a plane. The depth information mentioned earlier, which may be obtained by processing the light-field video stream and/or from another source, such as a depth sensor, may be used in the fitting of planes to the candidate targets 910.

In a step 660, a mask may be generated for each of the candidate targets 910, indicating which pixels within the candidate target 910 are suitable for use as target pixels. Target pixels may be those with the desired color/intensity characteristics for accurate recognition between frames.

In a step 670, some of the candidate targets may be selected as targets. This selection may be made, for example, based on whether each of the candidate targets 910 was readily and accurately mapped to a plane in the step 650, and/or whether each of the candidate targets 910 contains suitable target pixels, as determined in the step 660. The step 540 may then end 690.

FIG. 10 is a screenshot depicting selection of targets 1010 in the frame 700 of FIG. 7, according to one embodiment. As described above, the targets 1010 may be the candidate targets 910 that are readily approximated with planes and contain suitable target pixels.

Camera Pathway Generation from Targets

FIG. 11 is a flow diagram depicting the step 550 of generating the camera pathway from the method 500 of FIG. 5, according to one embodiment. The step 550 may utilize one or more of the following, which may be included in the light-field video stream and/or provided separately:

• • Depth maps for at least the key frames of the light-field video stream. As described above, the depth maps may be obtained by processing the light-field video stream and/or from one or more depth sensors.
  • Subaperture views for at least the key frames of the light-field video stream. For a plenoptic light-field camera, such as the camera 200, a subaperture view is an image generated from light-field data from the same location on each microlens of a microlens array, such as the microlens array 202. For a tiled camera array, a subaperture view is an image generated from light-field data from the same location on the image captured by each of the cameras of the tiled array. Subaperture views may be readily obtained for any frame of the light-field video stream by processing the light-field video stream, itself.
  • Camera-intrinsic parameters obtained from calibration of the light-field camera. Camera-intrinsic parameters may be unique to the light-field camera used to capture the light-field video stream.
  • Light-field optics parameters pertinent to one or more light-field optical elements of the light-field camera used to capture the light-field video stream. For example, the light-field optics parameters may include the distance between the microlens array 202 and the main lens 213, and the distance between the microlens array 202 and the image sensor 203.
  • Camera motion for at least a portion of the light-field video stream. The camera motion need not apply to the entire light-field video stream, but may rather be applicable to only a portion. In some embodiments, the camera motion may be obtained from data captured by other sensors. SLAM techniques or the like may be applied to such sensor data to obtain the camera motion.
  • The targets and target pixels identified in the step 540. These may, if desired, be supplemented with targets and/or target pixels selected by a user through the use of an input device, such as the user input 206 of the camera 200 of FIG. 2 or the user input 315 of the post-processing system 300 of FIG. 3.

The step 550 may be designed to provide output, which may include one or more of the following:

• • The camera pathway for the entire light-field video stream. The camera pathway may advantageously be provided with six degrees of freedom, as mentioned previously.
  • 3D mapping of targets and/or target pixels. If desired, the targets may be modeled in a virtual 3D scene, and the camera pathway may be generated relative to the virtual 3D scene.

The step 550 may track the 3D movement of the light-field camera with accuracy sufficient to enable visually precise insertion of computer-generated content into the light-field video stream. As part of the step 550, the motion of the light-field camera may be tracked with six degrees of freedom, and the targets may be mapped in 3D space. Depth mapping may be carried out as a necessary by-product of generation of the camera pathway.

As shown, the step 550 may begin 1110 with a step 1120 in which the light-field video stream is divided into sequences. Each sequence may begin with one of the key frames identified in the step 540.

In a step 1130, one of the sequences may be selected. For the first iteration, this may be the first sequence of the light-field video stream. The targets and target pixels of the first key frame may already have been selected in the step 540.

In a step 1140, the position and/or orientation of the targets may be tracked, in each frame of the sequence. In a step 1150, the position and/or orientation of the targets may be compared between frames of the sequence to obtain a portion of the camera pathway corresponding to that sequence. This may be done, for example, by comparing each pair of adjacent frames, modeling the position and/or orientation of each target for the new frame, and building the camera pathway for the new frame. Thus, the step 1140 and the step 1150 may be carried out synchronously.

Thus, the 3D model (map) of the targets and the camera pathway may be propagated from the key frame to the last frame of the sequence, which may be the key frame of the next sequence. Accordingly, the camera pathway may be generated one frame at a time until the portion of the camera pathway for that sequence is complete. At the end of the sequence, in a step 1160, the position and/or orientation of the target pixels in the key frame at the beginning of the next sequence may be obtained.

In a query 1170, a determination may be made as to whether the camera pathway has been generated for all sequences designated in the step 1120. If not, the system may return to the step 1130 and select the next sequence in the light-field video stream. The step 1140, the step 1150, and the step 1160 may be repeated until the query 1170 is answered in the affirmative. The step 550 may then end 1190.

If desired, user input may be gathered at any point in the performance of the step 550. For example, the user may help identify new targets and/or target pixels, confirm whether the 3D model of targets and/or target pixels is correct, and/or confirm whether each new portion of the camera pathway is correct. Thus, propagation of errors through the process may be avoided.

Image Stabilization

As described in connection with FIG. 5, in the step 560, the camera pathway obtained in the step 550 may be adjusted, for example, to provide image stabilization. Specifically, the camera pose may be stabilized through all frames via splines or other analytical solutions. The multi-view (4D) nature of the light-field may allow for adjusting the projection coordinates of the individual frames. This may be done by adjusting the U, V coordinates within each light-field frame so that frame-to-frame motion is smooth and continuous.

By using 4D data, parallax and image resolution can be maintained, avoiding the losses inherent in known image stabilization methods for 2D images. The limits of perspective shift may be governed by the specifications of the light-field optics. By generating an adjusted camera pathway in 3D space, using the camera pathway, a new sample from the 4D light-field can be produced, thus generating a near parallax-perfect camera move.

FIG. 12 is an illustration 1200 depicting the relationship between camera acceleration curves and light-field optical constraints, according to one embodiment. As shown, the light-field camera 200 may be used to capture a light-field video stream including the frame 700 of FIG. 7. The light-field camera 200 may be dollied away from the scene, as indicated by the arrow 1210. If the motion of the light-field camera 200 is not smooth, the camera 200 may follow a camera pathway 1220 with an erratic, jittery shape. This may adversely impact the quality of the video projected from the light-field video stream.

The camera pathway 1220 may be obtained with relatively high accuracy through use of the methods provided herein. Then, the camera pathway 1220 may be adjusted (for example, by using splines or the like), to generate the adjusted camera pathway 1230, which is much smoother. The configuration and/or positioning of the light-field optics within the light-field camera 200, such as the main lens 213, the microlens array 202, and the image sensor 203, may determine the size of the perspective limits 1240, within which the Center of Perspective of the light-field video stream may be adjusted for each frame.

A much smoother video stream may be projected from the light-field video stream, from the viewpoint of the adjusted camera pathway 1230. This video stream may be outputted to a display screen or the like for viewing.

Integration of Other Sensors

As mentioned previously, other sensors may be used to enable still more accurate generation of the camera pathway 1220. For example, camera position and/or orientation data derived from such sensors may be compared with that of the camera pathway 1220 computed by 3D mapping the targets and/or target pixels in 3D space. If desired, such sensor data may be used for each sequence, or even each frame-by-frame progression, of the step 540.

FIG. 13 is a screenshot 1300 depicting the use of LiDAR and gyroscope data in combination with a light-field video stream to generate more accurate camera pathways. Specifically, the screenshot 1300 depicts the camera pathway 1220, which may be computed by 3D mapping the targets and/or target pixels in 3D space, as described previously. The camera pathway 1220 generated in this way may be compared with a corresponding camera pathway 1310 generated through use of LiDAR data, and/or a corresponding camera pathway 1320 generated through the use data from a gyroscope mounted on the camera 200.

Such comparison may be performed manually by a user, or automatically by the computing device. The camera pathway 1220 may, if desired, be modified based on the corresponding camera pathway 1310 and/or the corresponding camera pathway 1320. Such modification may also be carried out manually or automatically, and may be done in the course of performance of the step 540. In the alternative, distinct camera pathways may be computed in their entirety, and then compared and/or modified after the step 540 is complete.

The above description and referenced drawings set forth particular details with respect to possible embodiments. Those of skill in the art will appreciate that the techniques described herein may be practiced in other embodiments. First, the particular naming of the components, capitalization of terms, the attributes, data structures, or any other programming or structural aspect is not mandatory or significant, and the mechanisms that implement the techniques described herein may have different names, formats, or protocols. Further, the system may be implemented via a combination of hardware and software, as described, or entirely in hardware elements, or entirely in software elements. Also, the particular division of functionality between the various system components described herein is merely exemplary, and not mandatory; functions performed by a single system component may instead be performed by multiple components, and functions performed by multiple components may instead be performed by a single component.

Reference in the specification to “one embodiment” or to “an embodiment” means that a particular feature, structure, or characteristic described in connection with the embodiments is included in at least one embodiment. The appearances of the phrase “in one embodiment” in various places in the specification are not necessarily all referring to the same embodiment.

Some embodiments may include a system or a method for performing the above-described techniques, either singly or in any combination. Other embodiments may include a computer program product comprising a non-transitory computer-readable storage medium and computer program code, encoded on the medium, for causing a processor in a computing device or other electronic device to perform the above-described techniques.

Some portions of the above are presented in terms of algorithms and symbolic representations of operations on data bits within a memory of a computing device. These algorithmic descriptions and representations are the means used by those skilled in the data processing arts to most effectively convey the substance of their work to others skilled in the art. An algorithm is here, and generally, conceived to be a self-consistent sequence of steps (instructions) leading to a desired result. The steps are those requiring physical manipulations of physical quantities. Usually, though not necessarily, these quantities take the form of electrical, magnetic or optical signals capable of being stored, transferred, combined, compared and otherwise manipulated. It is convenient at times, principally for reasons of common usage, to refer to these signals as bits, values, elements, symbols, characters, terms, numbers, or the like. Furthermore, it is also convenient at times, to refer to certain arrangements of steps requiring physical manipulations of physical quantities as modules or code devices, without loss of generality.

It should be borne in mind, however, that all of these and similar terms are to be associated with the appropriate physical quantities and are merely convenient labels applied to these quantities. Unless specifically stated otherwise as apparent from the following discussion, it is appreciated that throughout the description, discussions utilizing terms such as “processing” or “computing” or “calculating” or “displaying” or “determining” or the like, refer to the action and processes of a computer system, or similar electronic computing module and/or device, that manipulates and transforms data represented as physical (electronic) quantities within the computer system memories or registers or other such information storage, transmission or display devices.

Certain aspects include process steps and instructions described herein in the form of an algorithm. It should be noted that the process steps and instructions of described herein can be embodied in software, firmware and/or hardware, and when embodied in software, can be downloaded to reside on and be operated from different platforms used by a variety of operating systems.

Some embodiments relate to an apparatus for performing the operations described herein. This apparatus may be specially constructed for the required purposes, or it may comprise a general-purpose computing device selectively activated or reconfigured by a computer program stored in the computing device. Such a computer program may be stored in a computer readable storage medium, such as, but is not limited to, any type of disk including floppy disks, optical disks, CD-ROMs, magnetic-optical disks, read-only memories (ROMs), random access memories (RAMs), EPROMs, EEPROMs, flash memory, solid state drives, magnetic or optical cards, application specific integrated circuits (ASICs), and/or any type of media suitable for storing electronic instructions, and each coupled to a computer system bus. Further, the computing devices referred to herein may include a single processor or may be architectures employing multiple processor designs for increased computing capability.

The algorithms and displays presented herein are not inherently related to any particular computing device, virtualized system, or other apparatus. Various general-purpose systems may also be used with programs in accordance with the teachings herein, or it may prove convenient to construct more specialized apparatus to perform the required method steps. The required structure for a variety of these systems will be apparent from the description provided herein. In addition, the techniques set forth herein are not described with reference to any particular programming language. It will be appreciated that a variety of programming languages may be used to implement the techniques described herein, and any references above to specific languages are provided for illustrative purposes only.

Accordingly, in various embodiments, the techniques described herein can be implemented as software, hardware, and/or other elements for controlling a computer system, computing device, or other electronic device, or any combination or plurality thereof. Such an electronic device can include, for example, a processor, an input device (such as a keyboard, mouse, touchpad, trackpad, joystick, trackball, microphone, and/or any combination thereof), an output device (such as a screen, speaker, and/or the like), memory, long-term storage (such as magnetic storage, optical storage, and/or the like), and/or network connectivity, according to techniques that are well known in the art. Such an electronic device may be portable or nonportable. Examples of electronic devices that may be used for implementing the techniques described herein include: a mobile phone, personal digital assistant, smartphone, kiosk, server computer, enterprise computing device, desktop computer, laptop computer, tablet computer, consumer electronic device, television, set-top box, or the like. An electronic device for implementing the techniques described herein may use any operating system such as, for example: Linux; Microsoft Windows, available from Microsoft Corporation of Redmond, Wash.; Mac OS X, available from Apple Inc. of Cupertino, Calif.; iOS, available from Apple Inc. of Cupertino, Calif.; Android, available from Google, Inc. of Mountain View, Calif.; and/or any other operating system that is adapted for use on the device.

In various embodiments, the techniques described herein can be implemented in a distributed processing environment, networked computing environment, or web-based computing environment. Elements can be implemented on client computing devices, servers, routers, and/or other network or non-network components. In some embodiments, the techniques described herein are implemented using a client/server architecture, wherein some components are implemented on one or more client computing devices and other components are implemented on one or more servers. In one embodiment, in the course of implementing the techniques of the present disclosure, client(s) request content from server(s), and server(s) return content in response to the requests. A browser may be installed at the client computing device for enabling such requests and responses, and for providing a user interface by which the user can initiate and control such interactions and view the presented content.

Any or all of the network components for implementing the described technology may, in some embodiments, be communicatively coupled with one another using any suitable electronic network, whether wired or wireless or any combination thereof, and using any suitable protocols for enabling such communication. One example of such a network is the Internet, although the techniques described herein can be implemented using other networks as well.

While a limited number of embodiments has been described herein, those skilled in the art, having benefit of the above description, will appreciate that other embodiments may be devised which do not depart from the scope of the claims. In addition, it should be noted that the language used in the specification has been principally selected for readability and instructional purposes, and may not have been selected to delineate or circumscribe the inventive subject matter. Accordingly, the disclosure is intended to be illustrative, but not limiting.

Claims

1. A method for processing a light-field video stream, the method comprising: at a processor, receiving a light-field video stream generated by a light-field camera;at the processor, selecting a plurality of target pixels in a plurality of key frames comprising at least a first frame and a second frame of the light-field video stream, wherein selecting the plurality of target pixels comprises selecting as target pixels those pixels having at least one of a predetermined color characteristic or a predetermining intensity characteristic and are selected from planar regions of static, textured objects appearing in the key frames of the light-field video stream as the target pixels, and wherein selecting the plurality of target pixels further comprises: generating a list of a plurality of targets appearing in at least one of the key frames;generating a plane model of each of the targets for each of the key frames; andgenerating a mask for each of the targets for each of the key frames, indicating one or more target pixels within each of the targets;at the processor, using the target pixels to generate, in three dimensions, a camera pathway indicative of motion of the light-field camera during generation of the light-field video stream; andat the processor, using the generated camera pathway to process the light-field video stream.

2. The method of claim 1, wherein selecting the plurality of target pixels further comprises: performing superpixel segmentation of each of the key frames to identify superpixels;calculating a motion error map for each of the key frames;using the superpixels and motion error maps to assess texture and motion error for each of the superpixels of each of the key frames to identify a plurality of the superpixels as candidate targets;fitting a plane to each of the candidate targets for each of the key frames; andselecting the targets from among the candidate targets.

3. The method of claim 1, wherein using the generated camera pathway to process the light-field video stream comprises: adjusting the camera pathway to generate an adjusted camera pathway; andprojecting the light-field video stream to a viewpoint defined by the adjusted camera pathway to generate a projected video stream;wherein the method further comprises, at an output device, outputting the projected video stream.

4. A method for processing a light-field video stream, the method comprising: at a processor, receiving a light-field video stream generated by a light-field camera;at the processor, selecting a plurality of target pixels in a plurality of key frames comprising at least a first frame and a second frame of the light-field video stream, wherein selecting the plurality of target pixels comprises selecting as target pixels those pixels having at least one of a predetermined color characteristic or a predetermining intensity characteristic;at the processor, using the target pixels to generate, in three dimensions, a camera pathway indicative of motion of the light-field camera during generation of the light-field video stream;at the processor, using the generated camera pathway to process the light-field video stream; andwherein using the target pixels to generate the camera pathway comprises: receiving at least one selection from the group consisting of:camera-intrinsic parameters obtained from calibration of the light-field camera;light-field optics parameters pertinent to one or more light-field optical elements of the light-field camera; andcamera motion, generated by a sensor operating contemporaneously with capture of the light-field video stream, indicative of motion of the light-field camera during at least a segment of the light-field video stream; andusing the selection to generate the camera pathway.

5. The method of claim 4, wherein using the target pixels to generate the camera pathway comprises: generating the camera pathway with six degrees of freedom for an entirety of the light-field video stream; andgenerating a 3D mapping of the target pixels.

6. A method for processing a light-field video stream, the method comprising: at a processor, receiving a light-field video stream generated by a light-field camera;at the processor, selecting a plurality of target pixels in a plurality of key frames comprising at least a first frame and a second frame of the light-field video stream;at the processor, using the target pixels to generate, in three dimensions, a camera pathway indicative of motion of the light-field camera during generation of the light-field video stream, wherein using the target pixels to generate the camera pathway comprises: dividing the light-field video stream into a plurality of sequences, each of which begins with one of the key frames;for each sequence: tracking a position and/or orientation of each of the target pixels in each frame;comparing the position and/or orientation of each of the target pixels between frames of the sequence to obtain a portion of the camera pathway for that sequence; andobtaining the position and/or orientation of each of the target pixels for the key frame for the next sequence;generating the camera pathway with six degrees of freedom for an entirety of the light-field video stream; andgenerating a 3D mapping of the target pixels; andat the processor, using the generated camera pathway to process the light-field video stream.

7. The method of claim 6, wherein using the generated camera pathway to process the light-field video stream comprises: adjusting the camera pathway to generate an adjusted camera pathway; andprojecting the light-field video stream to a viewpoint defined by the adjusted camera pathway to generate a projected video stream;wherein the method further comprises, at an output device, outputting the projected video stream.

8. The method of claim 7, wherein adjusting the camera pathway to generate the adjusted camera pathway comprises causing the adjusted camera pathway to be more stable than the camera pathway.

9. The method of claim 7, wherein adjusting the camera pathway to generate the adjusted camera pathway further comprises adjusting U, V coordinates within each of a plurality of frames of the light-field video stream to cause frame-to-frame motion to be relatively smooth and contiguous.

10. A non-transitory computer-readable medium for processing a light-field video stream, comprising instructions stored thereon, that when executed by one or more processors, perform the steps of: receiving a light-field video stream generated by a light-field camera;selecting a plurality of target pixels in a plurality of key frames comprising at least a first frame and a second frame of the light-field video stream, wherein selecting the plurality of target pixels comprises: selecting pixels with predetermined color and/or intensity characteristics by selecting pixels from planar regions of static, textured objects appearing in the key frames of the light-field video stream as the target pixels by: generating a list of a plurality of targets appearing in at least one of the key frames;generating a plane model of each of the targets for each of the key frames; andgenerating a mask for each of the targets for each of the key frames, indicating one or more target pixels within each of the targets;using the target pixels to generate, in three dimensions, a camera pathway indicative of motion of the light-field camera during generation of the light-field video stream; andusing the generated camera pathway to process the light-field video stream.

11. The non-transitory computer-readable medium of claim 10, wherein selecting the plurality of target pixels further comprises: performing superpixel segmentation of each of the key frames to identify superpixels;calculating a motion error map for each of the key frames;using the superpixels and motion error maps to assess texture and motion error for each of the superpixels of each of the key frames to identify a plurality of the superpixels as candidate targets;fitting a plane to each of the candidate targets for each of the key frames; andselecting the targets from among the candidate targets.

12. The non-transitory computer-readable medium of claim 10, wherein using the generated camera pathway to process the light-field video stream comprises: adjusting the camera pathway to generate an adjusted camera pathway such that the adjusted camera pathway is more stable than the camera pathway;projecting the light-field video stream to a viewpoint defined by the adjusted camera pathway to generate a projected video stream;wherein the non-transitory computer-readable medium further comprises instructions stored thereon, that when executed by one or more processors, cause an output device to output the projected video stream.

13. A system for processing a light-field video stream, the system comprising: a processor configured to: receive a light-field video stream generated by a light-field camera;select a plurality of target pixels in a plurality of key frames comprising at least a first frame and a second frame of the light-field video stream by: selecting pixels from planar regions of static, textured objects appearing in the key frames of the light-field video stream as the target pixels;generating a list of a plurality of targets appearing in at least one of the key frames;generating a plane model of each of the targets for each of the key frames; andgenerating a mask for each of the targets for each of the key frames, indicating one or more target pixels within each of the targets;use the target pixels to generate, in three dimensions, a camera pathway indicative of motion of the light-field camera during generation of the light-field video stream; anduse the generated camera pathway to process the light-field video stream; andan output device configured to output the light-field video stream.

14. The system of claim 13, wherein the processor is further configured to select the plurality of target pixels by: performing superpixel segmentation of each of the key frames to identify superpixels;calculating a motion error map for each of the key frames;using the superpixels and motion error maps to assess texture and motion error for each of the superpixels of each of the key frames to identify a plurality of the superpixels as candidate targets;fitting a plane to each of the candidate targets for each of the key frames; andselecting the targets from among the candidate targets.

15. The system of claim 13, wherein the processor is further configured to use the target pixels to generate the camera pathway by: receiving at least one selection from the group consisting of: camera-intrinsic parameters obtained from calibration of the light-field camera;light-field optics parameters pertinent to one or more light-field optical elements of the light-field camera; andcamera motion, generated by a sensor operating contemporaneously with capture of the light-field video stream, indicative of motion of the light-field camera during at least a segment of the light-field video stream; andusing the selection to generate the camera pathway.

16. The system of claim 13, wherein the processor is further configured to use the generated camera pathway to process the light-field video stream by: adjusting the camera pathway to generate an adjusted camera pathway such that the adjusted camera pathway is more stable than the camera pathway; andprojecting the light-field video stream to a viewpoint defined by the adjusted camera pathway to generate a projected video stream;and wherein the output device is further configured to output the projected video stream.

17. A system for processing a light-field video stream, the system comprising: a processor configured to: receive a light-field video stream generated by a light-field camera;select a plurality of target pixels in a plurality of key frames comprising at least a first frame and a second frame of the light-field video stream;use the target pixels to generate, in three dimensions, a camera pathway with six degrees of freedom and indicative of motion of the light-field camera during generation of the light-field video stream by: dividing the light-field video stream into a plurality of sequences, each of which begins with one of the key frames; andfor each sequence: tracking a position and/or orientation of each of the target pixels in each frame;comparing the position and/or orientation of each of the target pixels between frames of the sequence to obtain a portion of the camera pathway for that sequence; andobtaining the position and/or orientation of each of the target pixels for the key frame for the next sequence; andgenerating a 3D mapping of the target pixels;use the generated camera pathway to process the light-field video stream; andan output device configured to output the light-field video stream.

18. The system of claim 17, wherein the processor is further configured to use the generated camera pathway to process the light-field video stream by: adjusting the camera pathway to generate an adjusted camera pathway such that the adjusted camera pathway is more stable than the camera pathway; andprojecting the light-field video stream to a viewpoint defined by the adjusted camera pathway to generate a projected video stream;and wherein the output device is further configured to output the projected video stream.