大谷淳,岸野文郎
熱画像と可視光画像のステレオ統合による時系列距離画像獲得法の検討
Abstract:We propose a new method for acquiring time-sequential range images that provides dense range data in a viewpoint-independent way using only passive sensors. In our approach, stereo pairs of thermal and intensity images are synchronously acquired and are mutually registered. Since the points to be matched in the stereo thermal images are limited to those with the same temperature values, the thermal images are segmented into isotemperature regions whose contours are matched. Our method also exploits temporal correspondences between contours at successive time instants. Contour-based matching is done for the isotemperature regions independently at each time instant, and by temporal correspondence, possible matching pairs of contours are generated. By evaluating the similarities of the pairs, consistent and likely pairs are chosen. To supplement the sparse range data obtained from the contour matching process, information from the stereo intensity images is used. Intensity profiles along horizontal segments are matched by dynamic programming, where a horizontal segment is the intersection of an epipolar line with an isotemperature region. From corresponding pixel pairs obtained from the matching processes, the 3D coordinates of the points can be calculated. Experiments on real scenes, including a sequence showing moving human being, show promising results.