Symmetry Set And Its Use In Shape Recognition

Computer vision is often referred to as machine vision, is an interdisciplinary subject related many science fields, which research the world around us by image and video data. With the age of information, people begin to use computer more and more. The use of computer to deal with vision is become more and more important in the information of our country.Shape recognition is an essential and important part of computer vision, it is used in many field, such as image analysis, computer vision, and target recognition. While people can recognize objects easily, it is difficult for computer to recognize shapes quickly and accurately, especially when the shape is translated with rotation, translation, and scale transform. In this paper, we focus on two methods to study shape descriptor.Firstly, we present a novel V-system descriptor of a 2D model, which is invariant in the rotation, translation, and scale transform, removing the effect of the start point of the shape through the main orientation. We show that V-system Descriptor can be computed efficiently by the fast V-system transformation, and be good use of Multi Resolution Analysis through weighted Euclidean distance, so the new V-system Descriptor has some resistance to slightly block, and may achieve a good recognition according to the actual situation.Second, A novel shape descriptor is introduced, which is closely related to the Symmetry Set and the anti-Symmetry Set. A diagram with binary regions is obtained by pairs of points which share a geometrical property based on their mutual symmetry. The coefficient which is the representation of the upper triangular matrix of the diagram on the Walsh Function System on triangle domain is defined for the shape descriptor. This descriptor is tested with objects in different classes, and some objects are occluded or noised. The results are promising, the descriptor is invariant in the rotation, translation, and scale transformation, removing the effect of the start point of the shape, especially has some resistance to slight occlusion.