Computer vision makes machines to perceive circumstance with intelligence. It is important in both theoretical researches and practical applications. Most current computer vision algorithms are based on Marr’s vision theory. It lacks robustness dealing with complex conditions and is limited to work in the Euclidean space, which is different from the human vision system. This project originates from the study to human’s visual perception. Its major research issues include: Mathematical basis and theoretical framework of non-Euclidean space; Visual computing and pattern recognition of non-Euclidean space; Fusion of machine and human visual systems to explore cognitive mechanism of non-Euclidean space; Applications of non-Euclidean space in image and video analysis, object detection and recognition. This project attempts to change the basis of computer vision, i.e., the space it works in. It is promising to create a new direction of computer vision and to provide theories for practical applications such as multimedia retrieve, video surveillance and biometrics.
(Sources of Funding: NSFC)