Object recognition and segmentation via shape models

Altınoklu, Metin Burak
In this thesis, the problem of object detection, recognition and segmentation in computer vision is addressed with shape based methods. An efficient object detection method based on a sparse skeleton has been proposed. The proposed method is an improved chamfer template matching method for recognition of articulated objects. Using a probabilistic graphical model structure, shape variation is represented in a skeletal shape model, where nodes correspond to parts consisting of lines and edges correspond to pairwise relation between parts. For edge support function of lines, directional chamfer matching cost is calculated. The performance of the new method has been evaluated with experiments using databases especially suitable for shape based object detection methods. The proposed method performs well, and it is much faster as compared to related methods.