Geometric Computer Vision

Geometry is an important aspect of computer vision. The laws of geometry and optics describe how the three-dimensional world is imaged on the camera sensor and, hence, an understanding of imaging geometry is important for the development of automatic image analysis methods. The basis for this understanding is the geometry of multiple views, which we have studied from both theoretical and computational viewpoints. Also, geometric camera calibration is an important topic because it is a prerequisite for image-based metrology, and we have a strong expertise in that field. Another classical problem in geometric computer vision is the recovery of three-dimensional scene structure from multiple images, which is illustrated below. This problem area includes the structure from motion problem, which is the simultaneous recovery of sparse scene structure and camera motion from multiple views, and the multi-view stereo problem, which deals with dense image-based surface reconstruction given known camera motion. Besides measurement and modeling of rigid scenes captured with conventional photographic cameras, the research has recently expanded into other areas as well, such as calibration of non-classical cameras, analysis of dynamic and non-rigid scenes, and novel bioimaging applications.

quasidense_v2.png

Figure 1: Three-dimensional scene structure is represented by a quasi-dense point cloud recovered from three images.

In summary, we have a broad expertise in various problems of geometric computer vision. Our work covers both basic research and applications, and the main research themes are listed in the following. More details and information can be found from the links below.

CMV/Research/GeometricComputerVision (last edited 2011-11-19 14:09:22 by WebMaster)