Research And Application Of Plane Curve Matching Method

Shape matching has an important significance in the research of computer vision, pattern recognition and image analysis. In many areas it have a wide range of applications. Such as character recognition, visual navigation, detection, identification and tracking of objects, positioning satellite images, medical diagnostic and others. Shape is one of the most basic features of the objects, and the curve can represent the shape and structure of the object, so we put forward the curve matching. The study of2D curve matching has made some progress using Freeman chain. Based on the previous research, this paper proposes a new curve matching algorithm. The main work of this paper includes the following four aspects:1. This paper introduces the background and the meaning of the curve matching, and then it summarizes the current research status of the curve matching, include using the spline approximation to represent the curve and find the minimum matching error which based on the invariant features of curve, Finally summarizes the follow-up work such as splicing and restoration.2. Summarizes the current shape matching method, which mainly include representation curve, shape matching method and descriptions of the method. In actual operation, often using spline, chain code polygonal approximation and some feature points based on scale space represent. Shape matching method is divided into global invariant and local characteristics. The global invariant divided divided into the invariant feature based on transform domain and based on geometric features. Finally introduced several kinds of shape description method which based on the transform domain invariant such as Matrix, Fourier descriptor, wavelet descriptor, morphological descriptors, V descriptors.3. This paper put forward a two-dimensional curve matching algorithms based on the Freeman code, the main ideas of this algorithms are:first, get the graphics boundary curves through the boundary extraction algorithm, and get the Freeman codes by boundary tracking algorithm. Freeman code can be divided into two categories based on the change of Freeman code near the corner point, then, we use two judgment criteria to determine corner points. Some false corner points are removed by the length of the lines and the curvature of the feature points to get the standard corner points. Finally, Curves are matched together using the normalized cross correlation of the length and curvature. Compared to the previous method based on Freeman chain code, this new method may be implemented easily and not affected by rotation and translation, the real experiments show the effectiveness of this method.4. Summarizes the content of this paper, analysis the shortage and needs to be improved of the work in this paper, which could provide guidance and reference for future study and work.