Interactive imaging systems and methods for motion control by users

Abstract

In various embodiments, the present invention provides a system and associated methods of calibration and use for an interactive imaging environment based on the optimization of parameters used in various segmentation algorithm techniques. These methods address the challenge of automatically calibrating an interactive imaging system, so that it is capable of aligning human body motion, or the like, to a visual display. As such the present invention provides a system and method of automatically and rapidly aligning the motion of an object to a visual display.

Description

FIELD OF THE INVENTION

The present invention relates generally to the fields of interactive imaging and interactive imaging system calibration. More specifically, the present invention relates to an auto-calibrating interactive imaging system and a method by which the interactive imaging system is initialized and automatically calibrated by optimizing the parameters of a segmentation algorithm using an objective function.

BACKGROUND OF THE INVENTION

An interactive imaging experience includes an environment in which an interactive display is affected by the motion of human bodies, objects, or the like. A camera, or set of cameras, detects a number of features of the human bodies before the camera, such as their silhouettes, hands, head, and direction of motion, and determines how these features geometrically or photometrically relate to the visual display. For example, a user interacting before a front-projected display casts a shadow on an optional display medium such as a projection screen, or the like. The interactive imaging system is capable of aligning the camera's detection of the silhouette of the human body with the shadow of the human body. This geometric or photometric alignment creates a natural mapping for controlling elements in the visual display. Persons of all ages can likely recall an experience of playing with their shadows and can thus understand that their motion in front of a source of bright light will produce a shadow whose motion behaves exactly as expected. This experience is capitalized upon in an interactive imaging experience.

In order for interactive imaging systems to operate and function properly, such systems must be accurately calibrated and optimized first. Procedures exist under which the motion of the human body, or the like, is geometrically or photometrically aligned to the actual visual display, creating a natural mapping for use in an interactive imaging system. However, these interactive imaging devices and systems require an extensive period of time, often taking many hours, for calibration and initialization. Such a delay results in long periods of wait time with no use of the interactive imaging system upon setup, until such time the calibration period is completed. This is equivalent to powering on a personal computer, expecting to use it immediately, yet waiting for hours before actual use can begin. Thus, such methods of calibration in an interactive imaging system are not automatic and nearly instantaneous, as is desired.

Calibration in an interactive imaging system refers to the initialization and setting of various setup parameter values. These parameter values, once initialized, are used in various segmentation algorithms. Segmentation, generally, has to do with image processing. Segmentation is a technique concerned with splitting up an image, or visual display, into segments or regions, each segment or region holding properties distinct from the areas adjacent to it. This is often done using a binary mask, representing the presence of a foreground object in front of the visual display surface.

A conceptual example of this definition of segmentation is the image formed on an all-white front-projected visual display when a person, or the like, is placed in front of the visual display and casts a shadow upon it. In this example, only the black or shadowed region of the visual display, as viewed on a wall, projection screen, or the like, denotes the presence of a foreground element, a body or similar object, and the white color in the visual display denotes background or non-presence of a foreground object. Normally, however, this segmentation is a binary image representation that is computed using a monochrome camera input.

There are a number of segmentation techniques, or algorithms, which are already well-known in the art. Two of these segmentation techniques include background subtraction and stereo disparity-based foreground detection, both of which may be employed for generating a segmentation image.

All of these algorithms share the need to set parameters which affect the quality of the segmentation as defined by its similarity to ground truth and as defined by its speed of execution. Calibration is the process of setting these parameters in order to achieve high quality in a visual display while operating at an acceptable execution speed. Unfortunately, existing calibration methods in interactive imaging systems require too much time for actual calibration and optimization. Such time requirements produce unsuitable delays.

A common approach for generating segmentation images from a camera that faces a visual display is to filter the camera to observe only near-infrared light while ensuring that the display only emits visible, non-infrared light. By separating the sensing spectrum from the display spectrum, the problem is reduced from detecting foreground elements in a dynamic environment created by a changing display to the problem of detecting foreground elements in a static environment, similar to chroma-key compositing systems with green or blue screens.

Background subtraction is the most popular means of detecting foreground elements (segmentation) for real-time computer vision applications. A model of the background, B, is maintained over time and is usually represented as an image with no foreground elements. It is assumed that the camera can view the entire area covered by the visual display; however, it is not assumed that the boundaries of the camera align exactly with the boundaries of the visual display. Therefore, any image captured by the camera, including the background model, must be warped such that the boundaries of the visual display and warped image do align. Warping is performed by defining four coordinates in the camera image C₁, C₂, C₃, and C₄, and bilinearly interpolating the pixel values that are enclosed by a quadrilateral whose corners are defined by C₁, C₂, C₃, and C₄, As a result, the warped camera geometrically corresponds to the display. A method for automatically computing these coordinates in the camera using homographies was presented in R. Sukthankar, R. Stockton, M. Mullin. Smarter Presentations: Exploiting Homography in Camera-Projector Systems. Proceedings of International Conference on Computer Vision, 2001. (A homography is a 2D perspective transformation, represented by a 3×3 matrix that maps each pixel on a plane such as a camera's image plane to another plane, such as a projector's image plane, through an intermediate plane, such as the display surface.) This method, however, assumes that the display may be viewed by the camera and the camera whose image needs to be warped is infrared-pass filtered, therefore eliminating the visibility of the display. Additionally, an automatic camera-camera homography estimation method was disclosed by M. Brown and D. G. Lowe in Recognising Panoramas. In Proceedings of the 9th International Conference on Computer Vision (ICCV2003), pages 1218-1225, Nice, France, October 2003.

While these patents and other previous systems and methods have attempted to solve the above mentioned problems, none have provided an auto-calibrating interactive imaging system and a method by which the interactive imaging system is initialized and automatically calibrated by optimizing the parameters of a segmentation algorithm using an objective function. Thus, a need exists for a system and methods of calibration and use in an interactive imaging system in which the calibration of parameters for segmentation algorithms is completed at an acceptable execution speed, and in which there is no deterioration in the quality of the visual display images.

BRIEF SUMMARY OF THE INVENTION

In various embodiments, the present invention provides a system and methods of calibration and use for an interactive imaging environment based on various segmentation techniques. This system and associated methods address the challenge of automatically calibrating an interactive imaging system, so that it is capable of aligning human body motion, or the like, to a visual display. Although this disclosure details two segmentation algorithms that operate using specific hardware configurations, the disclosed calibration procedure, however, is general enough for use with other hardware configurations and segmentation algorithms.

The present invention addresses the challenge of automatically calibrating and optimizing an interactive imaging system, so that it is capable of aligning human body motion, or the like, to a visual display. As such the present invention is capable of automatically and rapidly aligning the motion of an object to a visual display.

In one exemplary embodiment of the present invention, an auto-calibrating interactive imaging system is disclosed. The auto-calibrating interactive imaging system includes a central control unit; an infrared image sensor; a visible image sensor; illumination energy devices, or the like, for illuminating the display surface with infrared light; a display of any kind, under the assumption that the display does not emit infrared light; and, optionally, a display medium.

In another exemplary embodiment of the present invention, a method of calibration and use in an interactive imaging system is provided in which the parameters for geometric calibration are automatically determined and initialized by optimizing an objective function. For example, using the background subtraction segmentation algorithm, the parameters to be optimized are C₁, C₂, C₃, and C₄, the warping parameters, which are coordinates in a camera image, corresponding to the corners of the projection.

In another exemplary embodiment of the present invention, a method of calibration and use in an interactive imaging system is provided in which the parameters for photometric calibration are automatically determined and initialized by optimizing an objective function. For example, using the background subtraction segmentation algorithm, the parameters to be optimized are a threshold, t, a median filter kernel, m, the number of median filter operations, n and the camera's exposure, e.

There has thus been outlined, rather broadly, the features of the present invention in order that the detailed description that follows may be better understood, and in order that the present contribution to the art may be better appreciated. There are additional features of the invention that will be described and which will form the subject matter of the claims. In this respect, before explaining at least one embodiment of the invention in detail, it is to be understood that the invention is not limited in its application to the details of construction and to the arrangements of the components set forth in the following description or illustrated in the drawings. The invention is capable of other embodiments and of being practiced and carried out in various ways. Also, it is to be understood that the phraseology and terminology employed are for the purpose of description and should not be regarded as limiting.

As such, those skilled in the art will appreciate that the conception, upon which this disclosure is based, may readily be utilized as a basis for the designing of other structures, methods, and systems for carrying out the several purposes of the present invention. It is important, therefore, that the claims be regarded as including such equivalent constructions insofar as they do not depart from the spirit and scope of the present invention.

Additional aspects and advantages of the present invention will be apparent from the following detailed description of an exemplary embodiment which is illustrated in the accompanying drawings.

BRIEF DESCRIPTION OF THE DRAWINGS

The present invention is illustrated and described herein with reference to various drawings, in which like reference numerals denote like apparatus components and/or method steps, and in which:

FIG. 1 is a schematic diagram illustrating the use of geometric calibration in an interactive imaging environment, particularly illustrating the projection image, according to an embodiment of the present invention.

FIG. 2 is a schematic diagram illustrating the use of geometric calibration in an interactive imaging environment, particularly illustrating the visible-pass filtered camera image, according to an embodiment of the present invention.

FIG. 3 is a schematic diagram illustrating the use of geometric calibration in an interactive imaging environment, particularly illustrating the infrared-pass filtered camera image, according to an embodiment of the present invention.

FIG. 4 is a schematic diagram illustrating the use of geometric calibration in an interactive imaging environment, particularly illustrating the various mapping techniques, according to an embodiment of the present invention.

FIG. 5 is a schematic diagram illustrating the use of photometric calibration in an interactive imaging environment, particularly illustrating the visual display goal, according to an embodiment of the present invention.

FIG. 6 is a schematic diagram illustrating the use of photometric calibration in an interactive imaging environment, particularly illustrating a non-optimized visual display with noise and having a poor visual quality, according to an embodiment of the present invention.

FIG. 7 is a schematic diagram illustrating the use of photometric calibration in an interactive imaging environment, particularly illustrating an optimized visual display with no noise and having a high visual quality, according to an embodiment of the present invention.

DETAILED DESCRIPTION OF THE INVENTION

Before describing the disclosed embodiments of the present invention in detail, it is to be understood that the invention is not limited in its application to the details of the particular arrangement shown since the invention is capable of other embodiments. Also, the terminology used herein is for the purpose of description and not of limitation.

In one exemplary embodiment of the present invention, a self-calibrating interactive imaging system 10 includes an image generator 20 operable for creating or projecting an image. The image generator 20 is, for example, a visible light projector or the like. Images that may be projected include, but are not limited to, calibration line-up silhouettes 60, waves, vapor trails, pool balls, etc. Optionally, the interactive imaging system 10 also includes a display medium 30 operable for receiving and displaying the created or projected image. The display medium 30 may include a two or three-dimensional projection screen, a wall or other flat surface, a television screen, a plasma screen, a rear-projection system, a hyper-bright organic light-emitting diode (OLED) surface (possibly sprayed-on as a flexible substrate and onto the surface of which images are digitally driven), or the like. In general, the interactive imaging system 10 is display agnostic.

The interactive imaging system 10 further includes one or more illumination energy devices 21 operable for flooding a field of view in front of the created or projected image with illumination energy. For example, the one or more illumination energy devices 21 may consist of one or more infrared lights operable for flooding the field of view in front of the created or projected image with infrared light of a wavelength of between about 700 nm and about 10,000 nm. Preferably, the infrared light consists of near-infrared light of a wavelength of between about 700 nm and about 1,100 nm. Optionally, the infrared light consists of structured (patterned) infrared light or structured (patterned) and strobed infrared light, produced via light-emitting diodes or the like. In an alternative exemplary embodiment of the present invention, the image generator 20 and the one or more illumination energy devices 21 are integrally formed and utilize a common illumination energy source.

The interactive imaging system 10 still further includes an infrared image sensor 24 operable for detecting the illumination energy which is in the infrared spectrum. The infrared image sensor 24 is, for example, an infrared-pass filtered camera, or the like. In an alternative exemplary embodiment of the present invention, the image generator 20 and the infrared image sensor 24 are integrally formed. Optionally, an optical filter is coupled with the infrared image sensor 24 and is operable for filtering out illumination energy, which is in the infrared spectrum, of a predetermined wavelength or wavelength range, such as, for example, visible light.

The interactive imaging system 10 still further includes a visible light image sensor 22 operable for detecting the illumination energy in the visible light spectrum. The visible light image sensor 22 is, for example, a visible-pass filtered camera, or the like. In an alternative exemplary embodiment of the present invention, the image generator 20 and the visible light image sensor 22 are integrally formed. In yet another alternative embodiment, the image generator 20, infrared image sensor 24, and the visible light image sensor 22 are integrally formed.

The interactive imaging system 10 still further includes a computer vision engine 23. The computer vision engine 23 is used to detect a calibration image, or line-up silhouette 60, and an actual body 62 input for purposes of calibrating the interactive imaging system 10. The computer vision engine 23 is operable for detecting one or more users, such as an actual body 62, in the field of view in front of the created or projected image and segmenting the actual body 62 and a background. The computer vision engine 23 gives the interactive imaging system 10 “sight” and provides an abstraction of the actual body 62 and the background. In this manner, the one or more actual body 62 and the background are separated and recognized. When properly implemented, the number of actual bodies 62 can be determined, even if there is overlap, and heads and hands may be tracked. Preferably, all of this takes place in real time, i.e. between about 1/60^th and 1/130^th of a second. Optionally, the computer vision engine 23 is operable for detecting an actual body 62 in the field of view in front of the created or projected image and segmenting the one or more actual body 62 and the background. The computer vision engine 23 further provides the control logic for calibrating the interactive imaging system 10 segmentation algorithms.

The interactive imaging system 10 still further includes a computer interaction engine 26 operable for inserting an abstraction related to the one or more actual body 62 and/or the background. The computer interaction engine 26 understands interactions between the one or more actual body 62 and/or the background and creates audio/visual signals in response to them. In this manner, the computer interaction engine 26 connects the computer vision engine 23 and a computer rendering engine 27 operable for modifying the created or projected image in response to the presence and/or motion of the one or more actual body 62, thereby providing user interaction with the created or projected image in a virtual environment. Again, all of this takes place in real time, i.e. between about 1/60^th and 1/130^th of a second.

The interactive imaging system 10 still further includes a central control unit 25 operable for controlling and coordinating the operation of all of the other components of the interactive imaging system 10. A central control unit 25 directly connects to the computer interaction engine 26, computer vision engine 23, computer rendering engine 27, visible light image sensor 22, infrared image sensor 24, image generator 20, and the illumination energy devices 21.

FIGS. 1 through 4 are illustrative of geometric calibration 12 examples in an interactive imaging system 10. FIGS. 5 through 7 are illustrative of photometric calibration 14 examples in an interactive imaging system 10.

Referring now to FIG. 1, a schematic diagram illustrating the use of geometric calibration 12 in an interactive imaging environment, particularly illustrating the projection image, is shown. A display medium 30, such as a projection screen, or the like, is illustrated with a line-up silhouette 60, resembling the outline of a human body, projected onto the display medium 30. As an interactive imaging system 10 user stands at a fixed location between the image generator 20 and the display medium 30, the actual body 62 presence is detected by both the visible image sensor 22 and the infrared image sensor 24. The only input on the part of the user or operator, represented as an actual body 62, is to stand in a fixed location with arms outspread, mimicking the line-up silhouette 60 projected onto the display medium 30. Each of the corners 32 of the display medium 30 are recognized by the computer vision engine 23 and central control unit 25 as P₁, P₂, P₃, and P₄.

Under certain circumstances, the display medium 30 is much larger in size than the actual body 62. For example, consider a twenty-five foot tall display medium 30. Although an actual body could stand nearer the image generator 20 and create a larger shadow, interactive imaging is better suited to the actual body that is within three to ten feet away from the display medium. In such an environment, the actual body could not cast a shadow large enough to fill the calibration image, the line-up silhouette 60. Fortunately, the central control computer 25 and the computer vision engine 23 will operate under a relative scale for calibration purposes. The actual user 62 can initiate the scaling-down process by beginning to slowly flap his or her arms up and down until the line-up silhouette has downsized (or upsized) to the appropriate scale for line up and calibration purposes.

Referring now to FIG. 2, a schematic diagram illustrating the use of geometric calibration 12 in an interactive imaging system 10, particularly illustrating the view as seen by the infrared image sensor 24, is shown. The image generator 20, visible image sensor 22, infrared image sensor 24, and interactive imaging system 10 are displayed. The infrared image sensor view 40 illustrates the view as it is seen by the infrared image sensor 24. The infrared image sensor view 40 shows the outer edges of the camera's viewing range. The features 42 visible in the infrared image sensor 24, such as the four corners of a screen, in this example, are shown. The features 42 visible in the infrared image sensor 24 need not be limited to just the four corners of a screen. The features 42 may be any four references points, including, but not limited to, posted notes, black dots, and visible ink. These features 42 also alternatively include other training or testing points, under the assumption that the points are located along the same plane, yet not in a linear arrangement. The features 42 visible in the infrared image sensor 24, are recognized by the central control unit 25 as S₁, S₂, S₃, and S₄.

Referring now to FIG. 3, a schematic diagram illustrating the use of geometric calibration 12 in an interactive imaging environment, particularly illustrating the visible-light image sensor view 50, is shown. The image generator 20, visible image sensor 22, infrared image sensor 24, and interactive imaging system 10 are displayed. The visible-light image sensor view 50 illustrates the view as it is seen by the visible image sensor 22. The visible-pass filtered camera view 50 shows the outer edges of the camera's viewing range. The features 52 visible in the visible light image sensor 22, such as the four corners of the display medium 32 in this example, are shown. Each of the corners 32 of the display medium 30 are recognized by the central control unit 25 as P₁, P₂, P₃, and P₄. Additionally, features 42 visible in the infrared-pass filtered camera 24 and features 52 visible in the visible-pass filtered camera 22 allow a mapping to be made from the infrared image sensor 24 to the visible image sensor 22. This mapping is illustrated as the features 42 visible in the infrared image sensor 24, or S₁, S₂, S₃, and S₄, as recognized by the central control unit 25, are shown in the visible-pass filtered camera view 50.

Referring now to FIG. 4, a schematic diagram illustrating the use of geometric calibration 12 in an interactive imaging environment is shown. Each of the views previously illustrating in FIGS. 1 through 3 are shown: the view projected by the image generator 20 to an actual body 62 and onto a display medium 30, the infrared image sensor view 40, and the visible light image sensor view 50. In addition, the interactive imaging system 10 is shown. Finally, the mappings 70, 72, 74 of the various images are shown to illustrate how geometric calibration 12 is completed in an interactive imaging system 10.

Features visible in the infrared image sensor 42 and features visible in the visible image sensor 52 allow for a mapping from the infrared image sensor 24 to the visible image sensor 22. Thus, mapping 1, IR-to-VIZ, 70 illustrates an infrared-to-visible homography. The infrared image sensor 24 is unable to view the image generator 20; however, the visible image sensor 22 is able to view the image generator 20. The ability to view the image generator 20 in the visible image sensor 22 allows a mapping to be made between the visible image sensor 22 to the image generator 20. Thus, mapping 2, VIZ-to-PROJ, 72 illustrates a visible-to-projector homography. Mapping, 3 IR-to-PROJ, 74 illustrates the multiplication of the results of mapping 2, VIZ-to-PROJ, 72 multiplied against the results of the mapping 1, IR-to-VIZ, 70. Mapping 3, IR-to-PROJ, 74 is a mapping from the infrared image sensor 24 to the image generator 20. Since the infrared image sensor 24 is unable to view the image generator 20, this mapping 3, IR-to-PROJ, 74 would not be possible without the use of the visible image sensor 22, which can see the image generator 20, and the intermediate mappings, Mapping 1, IR-to-VIZ, 70 and Mapping 2, VIZ-to-PROJ, 72.

Referring now to FIG. 5, a schematic diagram illustrating the use of photometric calibration 14 in an interactive imaging environment is shown. The image generator 20, visible image sensor 22, infrared image sensor 24, and interactive imaging system 10 are displayed. Illustrated on the display medium 30 is a line-up silhouette 60 which has been projected onto the display medium 30 by the image generator 20. The line-up silhouette 60 is a calibration image. An actual body 62 (not shown) is to stand before the image generator 20 and try to fit his or her shadow into the displayed line-up silhouette 60. As the actual body 62 remains in location, the interactive image system 10 will calibrate itself, initializing the appropriate parameters to use in a given segmentation algorithm for photometric calibration 14.

Referring now to FIG. 6, a schematic diagram illustrating the use of photometric calibration 14 in an interactive imaging environment is shown. This figure illustrates the projected image before photometric calibration 14 has taken place. The image generator 20, visible image sensor 22, infrared image sensor 24, and interactive imaging system 10 are displayed. No real person or actual body 62 is shown here. Illustrated on the display medium 30 is a line-up silhouette 60 (a calibration image), which has been projected onto the display medium 30 by the image generator 20. The noise 80 located sporadically across the display medium 30 illustrates an attempt by the image generator 20, the central control unit 25, and the computer vision engine 23 to project a segmented silhouette 64 (shown in FIG. 7). However, the high levels of noise 80 and poor visual quality is resultant when optimization of the interactive imaging system 10 has not occurred. The noise 80 is what the system senses as a representation of an actual body 62. FIG. 6 illustrates an interactive imaging system 10 that is clearly not calibrated and optimized.

Referring now to FIG. 7, a schematic diagram illustrating the use of photometric calibration 14 in an interactive imaging environment is shown. This figure illustrates the projected image after photometric calibration 14 has taken place. The image generator 20, visible image sensor 22, infrared image sensor 24, and interactive imaging system 10 are displayed. Shown on the display medium 30 are both line-up silhouette 60 and the segmented silhouette 64. The line-up silhouette 60 has been projected onto the display medium 30 by the image generator 20. An actual body 62 (not shown) is to stand before the image generator 20 and try to fit his or her shadow into the displayed line-up silhouette 60. As the photometric calibration 14 takes place, the image generator 20, central control unit 25, and the computer vision engine 23 calculate and project a segmented silhouette 64 onto the display medium. Thus, after the process of photometric calibration 14, during which noise is removed and the decision variables are optimized to reach the objective, there is no noise and a high visual display quality results. The segmented silhouette 64 matches very closely to the line-up silhouette after the photometric calibration 14 process. The photometric calibration 14 process includes attaching a score to the measured differences between the many x,y coordinates of line-up silhouette 60 and the segmented silhouette 64. Any pixel coordinates that are different are counted as a point. A given configuration of the decision variables results in the score. The objective is to find the set of parameters or decision variable assignments that result in the lowest score. The lower the resultant score, the closer the segmented silhouette 64 is to the line-up silhouette 60. Once they are matching, the interactive imaging system is calibrated.

A contribution of this system and method is to use a second camera, a visible pass filtered camera, to automatically estimate camera coordinates. This system and method combines the automatic projector-camera and homography estimation method of R. Sukthankar, R. Stockton, M. Mullin. Smarter Presentations: Exploiting Homography in Camera-Projector Systems. Proceedings of International Conference on Computer Vision, 2001 and the automatic camera-camera homography estimation method of M. Brown and D. G. Lowe. Recognising Panoramas. In Proceedings of the 9th International Conference on Computer Vision (ICCV2003), pages 1218-1225, Nice, France, October 2003.

A homography is a 2D perspective transformation, represented by a 3×3 matrix, that maps each pixel on a plane such as a camera's image plane to another plane, such as a projector's image plane, through an intermediate plane, such as the display surface. By computing a homography between two planes, we may look up the corresponding pixel locations between the two planes. A camera-projector homography, for example, would enable the determination of the location of a projector's corner (such as the origin coordinate at x=0, y=0) to the same location in the camera (such as x=13, y=47). By estimating the projectorvisible_camera homography and estimating the IR_cameravisible camera homography, one may find corresponding pixel locations between the projector and IR-pass camera. This enables the automatic determination of warping parameters C₁, C₂, C₃, C₄.

During segmentation runtime, each camera snapshot F is subtracted from the background model and the resulting difference image D=F−B is further processed to generate a binary segmentation output. A threshold variable t is used to evaluate D according to the following: If absolute_value(D)>t, output a white pixel denoting foreground, else output a black pixel denoting background. This result of this threshold operation, S, may be immediately used as a segmentation as it is a binary image with a (probably noisy) representation of the foreground. Following the threshold operation, a median filter is performed to eliminate small foreground connected components which may result from noise or error in the threshold setting. The number of median filter operations n and size of the median filter kernel m may be tuned to produce different results. Furthermore, the camera's exposure e may be changed to produce darker images if the image is overexposed and brighter images if underexposed.

The background subtraction technique for generating segmentation images requires setting the following parameters: C₁, C₂, C₃, C₄, t, m, n and e. C₁, C₂, C₃, C₄ are parameters for geometric calibration and t, m, n and e are photometric calibration parameters. The disclosed method is capable of automatically tuning these parameters by optimizing an objective function. The objective function evaluates the difference between the segmentation algorithm computed with given assigned values of the parameters or decision variables and ground truth or model of an expected segmentation for a given human configuration. The only input on the part of the user or operator is to stand in a fixed location with arms outspread, or another easily attainable, simple stationary pose.

The objective function that is optimized includes gradient descent (see Eric W. Weisstein. “Method of Steepest Descent.” From Math World—A Wolfram Web Resource. http://mathworld.wolfram.com/MethodofSteepestDescent.html), Levenberg-Marquardt, (see Eric W. Weisstein. “Levenberg-Marquardt Method.” From Math World—A Wolfram Web Resource. http://mathworld.wolfram.com/Levenberg-MarquardtMethod.html), and the like. Each is an optimization technique of applied mathematics and is well-known in the art.

In a preferred embodiment of the invention, the interactive imaging system 10 is set-up in an appropriate location and powered on. The image generator 20 projects a calibration image, a line-up silhouette 62, onto a display medium 30. As an interactive imaging system 10 user stands at a fixed location between the image generator 20 and the display medium 30, the actual body 62 presence is detected by both the visible image sensor 22 and the infrared image sensor 24.

Depending on which calibration method is used and depending on which segmentation algorithmic is used, various parameters will be set and initialized, and then optimized in an objective function. These parameter values, once initialized, are used in various segmentation algorithms. Calibration methods include, but are not limited to, geometric calibration 12 and photometric calibration 14. Segmentation algorithms or techniques include, but are not limited to, background subtraction and stereo disparity-based foreground detection.

For example, if geometric calibration and background subtraction are chosen, the parameters to be optimized are C₁, C₂, C₃, and C₄, the warping parameters, which are coordinates in a camera image. In such an example, the infrared image sensor 42 and the visible image sensor 52 are both viewing the display medium 30 and actual body 62 during calibration. Features visible in the infrared image sensor 42 and features visible in the visible image sensor 52 allow for a mapping from the infrared image sensor 24 to the visible image sensor 22. Mapping 1, IR-to-VIZ, 70 illustrates an infrared-to-visible homography. The infrared image sensor 24 is unable to view the image generator 20; however, the visible image sensor 22 is able to view the image generator 20. The ability to view the image generator 20 in the visible image sensor 22 allows a mapping to be made between the visible image sensor 22 to the image generator 20. Mapping 2, VIZ-to-PROJ, 72 illustrates a visible-to-projector homography. Mapping, 3 IR-to-PROJ, 74 illustrates the multiplication of the results of mapping 2, VIZ-to-PROJ, 72 multiplied against the results of the mapping 1, IR-to-VIZ, 70. Mapping 3, IR-to-PROJ, 74 is a mapping from the infrared image sensor 24 to the image generator 20.

The coordinates in the camera image, C₁, C₂, C₃, C₄ are the parameters for geometric calibration. As the parameter values are changed, various results are produced. By estimating the VIZ-to-PROJ homography and estimating the IR-to-VIZ homography, one may find corresponding pixel locations between the image generator 20 and the infrared image sensor 24. This enables the automatic determination of warping parameters C₁, C₂, C₃, C₄.

By incorporating the use of an objective function, the differences between the segmentation algorithm, computed with given assigned values of the parameters or decision variables, and ground truth, or model of an expected segmentation for a given actual body 62 configuration, are evaluated. This in effect mathematically determines the correctness or goodness of a parameter value. With the rapid optimization of an objective function, good parameter values can be quickly set and the segmented silhouette 64, with no noise and with a high visual quality is reached, thus calibrating the interactive imaging system 10.

Although the present invention has been illustrated and described with reference to preferred embodiments and examples thereof, it will be readily apparent to those of ordinary skill in the art that other embodiments and examples may perform similar functions and/or achieve similar results. All such equivalent embodiments and examples are within the spirit and scope of the invention and are intended to be covered by the following claims.

Claims

1. A system, comprising: an infrared image sensor configured to detect infrared illumination energy in a field of view;a light sensor configured to detect visible light in the field of view; anda control unit communicatively coupled to the infrared image sensor and the light sensor, wherein the control unit is configured to: detect interactions between a user and a display medium and create audio/visual signals in response;modify the audio/visual signals in response to the presence and/or motion of the user, thereby providing user interaction in a virtual environment; anduse the light sensor to capture an image of the user or other calibration object and display or project the image on the display medium.

2. The system of claim 1, wherein the control unit is configured to: calibrate the infrared image sensor and/or the light sensor using a segmentation algorithm with a plurality of parameters and an objective function to set and optimize the plurality of parameters of the segmentation algorithm.

3. The system of claim 2, wherein the objective function comprises a gradient descent method.

4. The system of claim 2, wherein the objective function comprises a Levenberg-Marquardt method.

5. The system of claim 1, further comprising: one or more illumination energy devices configured to flood the field of view in front of the projected image with illumination energy for detection by the infrared image sensor.

6. The system of claim 1, wherein the segmentation algorithm is background subtraction.

7. The system of claim 1, wherein the segmentation algorithm is stereo disparity-based foreground detection.

8. The system of claim 1, wherein the control unit is configured to: detect the user with both the infrared image sensor and the light sensor.

9. The system of claim 1, wherein the control unit is configured to: display an image of the user on the display medium.

10. A control unit, comprising: connections to an infrared image sensor configured to detect infrared illumination energy in a field of view and a light sensor configured to detect visible light in the field of view;a computer vision engine configured to display or project on a display medium a image of a user captured by the light sensor;a computer interaction engine configured to detect interactions between the user and images on the display medium and create audio/visual signals in response; anda computer rendering engine configured to modify the audio/visual signals in response to the presence and/or motion of the user, thereby providing user interaction in a virtual environment.

11. The control unit of claim 10, wherein the computer vision engine is configured to: calibrate the infrared image sensor and/or the light sensor using a segmentation algorithm with a plurality of parameters and an objective function to set and optimize the plurality of parameters of the segmentation algorithm.

12. The control unit of claim 11, wherein the objective function comprises a gradient descent method.

13. The control unit of claim 11, wherein the objective function comprises a Levenberg-Marquardt method.

14. The control unit of claim 10, wherein the connections further connect to one or more illumination energy devices configured to flood the field of view in front of the projected image with illumination energy for detection by the infrared image sensor.

15. The control unit of claim 10, wherein the segmentation algorithm is background subtraction.

16. The control unit of claim 10, wherein the segmentation algorithm is stereo disparity-based foreground detection.

17. The control unit of claim 10, wherein the control unit is configured to: detect the user with both the infrared image sensor and the light sensor.

18. The control unit of claim 10, wherein the control unit is configured to: display an image of the user on the display medium.

19. A method, comprising: providing an infrared image sensor configured to detect infrared illumination energy in a field of view, a light sensor configured to detect visible light in the field of view, and a control unit communicatively coupled to the infrared image sensor and the light sensor;detecting interactions between a user and a display medium and create audio/visual signals in response;modifying the audio/visual signals in response to the presence and/or motion of the user, thereby providing user interaction in a virtual environment; andusing the light sensor to capture an image of the user and displaying or projecting the image on the display medium.