National Pattern Matching Algorithm Research And Application

National decoration pattern,which has very complex and rich cultural connotations and contains national culture,is a symbol of the national spirit,and embodies the aesthetic value and emotional value of the nation.Its aesthetic value not only keep it decorative,practical and public welfare in pattern design and development,but also make the national patterns become different,diverse,rich in content and have profoundness of good.This article topic come from the project of the Beijing Municipal Science and Technology Commission.The main aim of the research is to solve the ethnic decoration image segmentation pattern matching problem,by using matching algorithm,we should find out and display the similar patterns image in the database of national patterns.These similar patterns provide inspiration for designers,while the designers take innovation by using national pattern image in modern design,so that the national culture can maintain sustained vitality.In this paper,according to national pattern characteristics,we select the shape context algorithm to match the national pattern,but the method is not completely suitable for the national decoration database.The main work of this paper is as follows:1.According to characteristics of the national patterns,which are of single color with few pixels and different sizes,but of strong shape features,choose SIFT and shape context as matching algorithm to match the national pattern images.2.As simple shape context matching algorithm's rate of accuracy is low in matching the.A skeleton feature and shape context descriptor based shape matching method is proposed.This method add skeleton feature points to the original shape contour feature points for matching,using the two kinds of features together to generate shape matching descriptor.The test results in traditional shape databases show that the improved shape context matching algorithm can improve the matching accuracy and do not spend more time.3.Shape context method is not a rotation invariant while the national patterns have four angles of rotation,this article proposed a method of extracting the pattern direction vector.By rotating and matching the pattern direction vector based on LBS,and the patterns are also rotated and matched.So the matching price is obtained.In the national pattern database,the experiment results show that the improved algorithm is invariant to the rotation of the four angles of the national pattern,and the matching accuracy is greatly improved by 57%,it can find similar pattern with rotation variance.At the same time,the algorithm can also be applied to the traditional shape database,and maintain the algorithm matching degree.