The present invention relates generally to wide-Field of View (FOV) imaging devices, and more particularly, but not exclusively, to dual resolution wide FOV imaging system which is capable of simultaneously capturing a large FOV and a small FOV inside the said large FOV with much higher resolution.
Real-time acquisition of high-resolution, wide field of view (FOV) and high dynamic range (HDR) images is essential for many military and civilian surveillance applications. For instance, there is an urgent need for an omnidirectional imaging system in many surveillance applications where the system, with sufficient resolution and frame rate, can monitor the activities in all directions simultaneously across a very large operating field (e.g. spherical or complimentary hemispherical coverage) while being able to rapidly zoom into one or multiple objects of interest for reliable identification and characterization of the objects. Such a sensor needs to provide both excellent situational awareness and adequate detail resolvability. This type of sensors, if available, can find myriads of applications in both military and commercial markets.
However, when designing an optical imaging system, finite sensor resolution and data bandwidth impose limits on the spatial resolution and FOV achievable in state-of-the-art imaging systems. There is a well-known inherent tradeoff between the FOV and the resolving power for most conventional imaging techniques with a fixed number of pixels: the wider the FOV, the lower the resolving power. Using the traditional cluster-based omnidirectional cameras as an example, in order to achieve a 1 arc minute (−300 micro-rad) angular resolution, it requires at least 50 small FOV cameras (e.g. FOV: 33°×25°) with a 5-Mega pixel sensor on each to cover a spherical field of 360°×360°, which results in a minimum of 250 Mega pixels to be captured, stored and transmitted for a single spherical panoramic image, barring any pixel loss and FOV overlap. To achieve an angular resolution of 2 arc seconds requires a prohibitive number of cameras in the order of thousands to cover a spherical field. As a result, the cost and size of a camera-cluster-based system will be unacceptable for many surveillance applications, not mentioning that clustering over thousands of high-resolution cameras imposes great challenges to the state-of-the-art data management and image processing technologies.
Foveation techniques can actively track and capture a region of interest with high resolution sensor without losing the imaging capability of the peripheral area, similar to the foveation properties of the human vision system. Various imaging systems have been developed to explore the potential of applying the foveation technique in imaging applications. For example, Sandini et al. developed a retina-like CMOS sensor with spatially-variant resolution to mimic the human retina (G. Sandini, P. Questa, D. Scheffer and A. Mannucci, “A Retina-like CMOS sensor and its applications,” Proceedings of IEEE Workshop on Sensor Array and Multichannel Signal Process. (2000), pp. 514-9). Martinez and Wick proposed to use a liquid crystal spatial light modulator to dynamically correct the aberrations at the foveated region inside a wide FOV of imaging system (T. Martinez, D. V. Wick and S. R. Restaino, “Foveated, wide field-of-view imaging system using a liquid crystal spatial light modulator,” Opt. Express 8, 555-60 (2001); D. V. Wick, T. Martinez, S. R. Restaino and B. R. Stone, “Foveated imaging demonstration,” Opt. Express 10, 60-5 (2002)). The aforementioned approaches use only single-sensor to capture both the peripheral region and the foveated region which limits the overall information throughput of the system. Alternatively, Hua and Liu proposed a dual-sensor approach to the foveation imaging technology where two separate sensors are used to capture the peripheral region and the foveated region (Hong Hua and Sheng Liu, “Dual-Sensor foveated imaging system,” APPLIED OPTICS, Vol. 47, No. 3, 317-327, 2008). Comparing with the single sensor approach, the dual sensor approach uses two different sensors which can be in different size and different resolution, which has potential to yield high information throughput with low-cost detectors. The main dis-advantage of the dual-sensor approach is that the system employs an a focal system structure which usually has a limited ability to achieve large peripheral FOV and often results in a bulky system.
The present invention concerns a dual-sensor wide-FOV foveated imaging technique which is capable of acquiring wide-FOV videos of a surrounding space in real time and simultaneously obtaining very high resolution, high-magnification foveated images of multiple targets at high frame rate. The wide-FOV video, with appropriate resolution and frame rate, enables the real-time capability of simultaneously observing a surrounding space for acquiring, detecting, and tracking imminent threats posed by objects, while the high resolution foveated videos are focused on multiple small portions of the wide FOV in real time with substantially higher resolution to enable crucial target recognition and characterization. The region of interest (RoI) of the foveated view can be steered in real time to any part of the wide FOV image. These capabilities are analogous to the searching, tracking, and foveation functions of the human visual system. By integrating the foveation capability into a wide-FOV imaging system, the present invention is able to capture a wide viewing field up to 360°×360° with high angular resolution.
The present invention typically contains two subsystems: the wide-FOV imaging subsystem and the foveated imaging subsystem; and two subsystems are integrated as one system, where two imaging subsystems share the same objective lens, which may result in a compact and lightweight system design. The stop in the foveated imaging subsystem is optically conjugate with the stop in the wide-FOV imaging subsystem through the beamsplitter. For the present invention, the wide-FOV imaging subsystem captures a wide view field while the foveated imaging subsystem captures one or a few selected portions of the said wide view field and yields very high resolution videos to enable accurate target recognition. Compared with state-of-the-art surveillance systems, the present invention has the advantages of being relatively low-cost, compact, low power consumption, low data bandwidth demand as well as uncompromised high performance in terms of FOV, resolution, and real-time acquisition.
The objective lens of the present invention may utilize the rotationally symmetric refractive optical elements to capture an umbrella-like FOV or utilize a curved mirror along with necessary rotationally symmetric refractive optical elements to capture a ring-like panoramic FOV. The scanning mirror of the present invention may be a dual-axis scanning mirror to sample the wide-FOV using two tilting motions or may be a single-axis scanning mirror to sample the wide-FOV using a combined motion of tilting and rotation.
In one aspect of the present invention, the exemplary system may integrate multiple wide-FOV foveated imaging units to achieve a FOV much larger than that of a single unit. The integrated system may or may not possess single viewpoint properties. When a single viewpoint property is desired, a multi-faceted mirror may be used to virtually co-locate the viewpoints of all the imaging units in the integrated system to a single viewpoint.
The foregoing summary and the following detailed description of exemplary embodiments of the present invention may be further understood when read in conjunction with the appended drawings, in which:
a and 2b schematically illustrate two types of motions of the scanning mirror used in accordance with the present invention.
The embodiments according to the present invention will be fully described with respect to the attached drawings. The descriptions are set forth in order to provide an understanding of the invention. However, it will be apparent that the invention can be practiced without these details. Furthermore, the present invention may be implemented in various forms. However, the embodiments of the present invention described below shall not be constructed as limited to the embodiments set forth herein. Rather, these embodiments, drawings and examples are illustrative and are meant to avoid obscuring the invention.
A primary embodiment of the present invention comprises, a foveated imaging system (100), capable of capturing a wide field of view image and a foveated image, where the foveated image is a controllable region of interest of the wide field of view image, the system comprising:
In some embodiments, the incoming light from the external scene passes through the objective lens (110) to the beamsplitter (120), where the beamsplitter (120) divides the light into the two optical paths, a wide field of view imaging path (125) and a foveated imaging path (135). In the wide field of view path, the light passes through the first stop (127) to the wide field of view imaging lens (130) along the wide field of view imaging path (125). The lens focuses the wide field of view image upon the wide field of view imaging sensor (140). On the foveated view imaging path, the light passes through the second stop (137) to the scanning mirror (150) along the foveated imaging path (135), where the scanning mirror (150) reflects a region of interest toward the foveated imaging lens (160) through the beam splitter (120), The foveated imaging lens (160) focuses the foveated image upon the foveated imaging sensor (170).
In some embodiments, the objective lens (110) is disposed on the front of the system. The beamsplitter (120) is disposed adjacent to objective lens receiving light from the objective lens. The beamsplitter (120) divides the light into the two optical paths, a wide field of view imaging path (125) and a foveated imaging path (135). The first stop (127) is in optical communication with the beamsplitter (120) along the wide field of view imaging path (125) and the second stop (137) is in optical communication with the beamsplitter (120) along the foveated imaging path (135). The scanning mirror (150) is disposed near or at the position of the second stop (137), where it receives light from the beamsplitter (120) along the foveated imaging path (135) and reflects the light back to the beamsplitter (120). The wide-field of view imaging lens (130) is disposed to face the first stop (127) along the wide field of view imaging path (125), where it receives light from the beamsplitter (120) through the first stop (127) along the wide field of view path (125). The foveated imaging lens (160) is disposed to face the beamsplitter (120), where it receives light from the beamsplitter (120) reflected from the scanning mirror (150) along the foveated imaging path (135). The wide.-field of view imaging sensor (140) is disposed to face the wide field of view imaging lens (130). The foveated imaging sensor (170) is disposed to face the foveated imaging lens (160). The two images are recorded by the sensors, a wide field of view image and a high resolution image of the region of interest within it.
In some embodiments, the objective lens (110) is located on the front of the system. The beam splitter (120) is located between the objective lens and the stop (137) facing the objective lens (110) and the scanning mirror (150) so that it receives light from the objective lens. The scanning mirror (150) is located behind the beam splitter, where it receives light from the foveated image path of the beamsplitter (120) and reflects it back to the beamsplitter (120). The wide-field of view imaging lens (130) faces the wide field of view image path of the beam splitter, while the foveated imaging lens (160) faces the foveated image optical path of the beam splitter (120). The wide-field of view imaging sensor (140) faces the wide-field-of-view imaging lens (130), and the foveated imaging sensor (170) is faces the foveated imaging lens (160).
In some embodiments, the incoming light from the external scene passes through the objective lens (110) to the beamsplitter; whereupon the beam splitter (120) transmits one copy of the light to the wide field of view lens (130) and a second copy of the light to the scanning mirror (150). The scanning mirror (150) reflects a region of interest back to the beam splitter (120), and the beam splitter reflects the light to the foveated imaging lens (160). Meanwhile, the wide field of view imaging lens (130) transmits the light in the wide field of view imaging path (125) to the wide field of view image sensor (140). The foveated imaging lens (160) transmits the light in the foveated imaging path (135) to the foveated imaging sensor (170). Thus the two images are recorded by the sensors, a wide field of view image and a high resolution image of the region of interest within it.
In this exemplary layout 100, the objective lens 110 may be a group of rotationally symmetric lenses to capture a continuous umbrella-like FOV, or near-hemispherical-shape FOV, or near-spherical-shape FOV, The objective lens 110 could also contain a curved mirror surface along with necessary rotational symmetric lenses to capture a ring-like panoramic FOV. The curved mirror could be a spherical mirror, a parabolic mirror, a hyperbolic mirror, a conical mirror, an elliptical mirror, or aspherical mirror with or without symmetry or alike. The imaging sensors 140 and 170 can be any light sensing device containing an array of light sensing units (pixels) that converts photons into electronic signals, including, but not limited to, a charge-couple device (COD), or a complementary metal-oxide-semiconductor (CMOS) or other type of light sensing devices. The scanning mirror 150 can be any type of fast moving mirror devices whose scanning motion can be electronically controlled, including, but not limited to, voice coil mirror, piezoelectric mirror, Micro-Electro-Mechanical System (MEMS) mirror or other type of scanning mirrors. The beamsplitter 120 could be in form of a cube or a plate and could be a non-polarized beamsplitter or a polarized beamsplitter. When a polarized beamsplitter is used, a quarter-wave plate may be used along with the beamsplitter to increase the light efficiency. The quarter-wave plate may be positioned in the space between the beamsplitter 120 and the stop 137. Additional polarizers may be used in both the foveated imaging path 135 and the wide-FOV imaging path 125 to reduce the crosstalk between two paths.
As one of its benefits, the present invention combines two imaging subsystems into one integrated system, where two imaging subsystem share the same objective lens, which may result in a compact and lightweight system. The stop 137 in the foveated imaging subsystem is optically conjugate with the stop 127 in the wide-FOV imaging subsystem through the beamsplitter 120. For the present invention, the wide-FOV imaging subsystem captures a wide view field while the foveated imaging subsystem captures one or a few selected portions of the said wide view field and yields very high resolution videos to enable accurate target recognition. Compared with state-of-the-art surveillance systems, the present invention has the advantages of being relatively low-cost, compact, low power consumption, low data bandwidth demand as well as uncompromised high performance in terms of FOV, resolution, and real-time acquisition.
In one aspect of the present invention, the scanning mirror may be a dual axis scanning unit 252 for continuously sampling the wide-FOV through tilting motions 253 and 254 along X and Y axes as illustrated in
Compared to the dual sensor approach in the prior arts, the present invention uses a regular imaging system structure where the optical stop is inside the imaging system with a group of lenses are in front of the stop and a group of lenses are behind the stop. The advantages of using the regular imaging system structure over the afocal system in the prior art are:
In another significant aspect, the present invention uses a pair of optical conjugated stops which are inside the imaging system and created through a beamsplitter and located in the wide field of view and foveated view optical paths, respectively. In the prior art, the stop is placed at the entrance to an afocal system, and the image of the stop created through the afocal system is on the other side of the afocal system.
The yet another significant aspect, in the prior art, the scanning mirror is controllable only through X and Y tilt axes. In the present invention the scanning mirror may also be configured to use an X or Y tilt and Z rotation instead.
In one aspect of the present invention related to
This application claims priority to U.S. Provisional Application No. 61/620,581, filed on Apr. 5, 2012, and U.S. Provisional Application No. 61/620,574, filed on Apr. 5, 2012, the disclosures of which are incorporated herein by reference in its entirety.
Number | Date | Country | |
---|---|---|---|
61620581 | Apr 2012 | US | |
61620574 | Apr 2012 | US |