Panoramic Vision Sensors, Theory, and Applications

Current cameras are poor imitations of the human eye and close descen­ dants in their design of ideas and a technology that are more than a century old. People in computer vision have traditionally used off-the-shelf cameras that were not meant for the uses they were intended for by these researchers: off-the-shelf cameras are designed to capture images to be printed on paper or looked at on a television screen, not for guiding robots or making 3D models of the environment or even surveilling a large area where very large field of views, high geometric and photometric accuracies are necessary. Quite a significant part of the efforts in computer vision has been targeted at overcoming algorithmically these problems. The authors of this book convince us that it is possible to abandon the traditional route of using standard cameras and to follow the path of designing new cameras explicitly for solving the tasks at hand in computer vision applications. This leads to different design concepts and allows to alleviate many of the difficulties encountered in the processing of the images taken with the "traditional" cameras.

1 Introduction.- 2 A Brief Historical Perspective on Panorama.- Section I: Catadioptric Panoramic Systems.- 3 Development of Low-Cost Compact Omnidirectional Vision Sensors.- 4 Single Viewpoint Catadioptric Cameras.- 5 Epipolar Geometry of Central Panoramic Catadioptric Cameras.- 6 Folded Catadioptric Cameras.- Section II: Panoramic Stereo Vision Systems.- 7 A Real-time Panoramic Stereo Imaging System and Its Applications.- 8 Panoramic Imaging with Horizontal Stereo.- 9 Panoramic Stereovision Sensor.- 10 Calibration of the Stereovision Panoramic Sensor.- 11 Matching Linear Stereoscopic Images.- Section III: Techniques for Generating Panoramic Images.- 12 Characterization of Errors in Compositing Cylindrical Panoramic Images.- 13 Construction of Panoramic Image Mosaics with Global and Local Alignment.- 14 Self-Calibration of Zooming Cameras from a Single Viewpoint.- 15 360 × 360 Mosaics: Regular and Stereoscopic.- 16 Mosaicing with Strips on Adaptive Manifolds.- Section IV: Applications.- 17 3D Environment Modeling from Multiple Cylindrical Panoramic Images.- 18 N-Ocular Stereo for Real-Time Human Tracking.- 19 Identifying and Localizing Robots with Omnidirectional Vision Sensors.- 20 Video Representation and Manipulations Using Mosaics.