ALGORITHMS FOR POINT OBJECT DETECTION FROM STEREO IMAGES
V.S.Kirichuk and A.K.Shakenov
Novosibirsk
Pages: 13-20
Abstract
Various approaches to improving the quality of dynamic point object detection by using stereo vision are considered. A triangulation algorithm with preliminary thresholding and an algorithm forming a set of marks over the whole stereopsis set are implemented and studied. The dependence of the detection probability on the false alarm level was experimentally obtained for the proposed algorithms for different signal-to-noise ratios. A theoretical dependence of the object amplitude on the false alarm level for the given detection probability is presented. The efficiency of using a priori information on the altitude curve of an object while its detecting is analyzed.