A Study Of Encoding And Retrieval National Patterns With Chain Code

As material life become more and more rich, the study of culture whichrepresents the uniquely wisdom wealth of human appears more and more profound,therefore the preservation and inheritance of culture in different forms got theattention of the researchers. Meanwhile, image processing, such as image recognition,image matching and retrieval and so on, is a very important research field in computerscience. In this paper, the Chinese national patterns, the study object of our work, isan significant part of national culture and image processing.Many of national patterns forms in different kind of lines, it can be saved andwell showed in vector. An image represented in vector is made up of separate functioncurve which combined freely and unlimited, and it will not be distortion whenenlarged but it is not suitable for search. In this paper, we proposed a new method totransform the Chinese national patterns form vector to chain code according thedifferent character of different kinds of lines. The encoding part includes two parts:one shows the structure characteristics of the patterns, the other shows the directioncharacteristics through encoding the feature point of each line by Freeman chain code.The storage of the chain code gets a number of compression compared to the bitmapformat. It shows that the chain code format can not only well express the imagecontour but also well works on complex linear pattern with intersections andbreakpoints.A new retrieval method of Chinese national patterns is proposed based on thedirection information and structure information showed in the new encoding methodin this paper. The definition Direction chain code entropy (DCCE) is put forward bycombining the information entropy in informatics which used to express the amountof information in an information system with the chain code encoding. The new retrieval method using DCCE and structure information can effectively match thesimilar patterns of target pattern in the nation patterns library.Experiment process is divided into two parts, the first part is the line comparisonexperiment, according to different characteristics of the lines in different forms, theexperiment results prove that the proposed encoding scheme can effectivelydistinguish the obvious differences between lines in same type. The second part,based on the national patterns library, the experiment result verified the new retrievalmethod this paper puts forward can effectively match the similar patterns of targetpattern in the nation patterns library. At the same time we compared our method withcommonly used chain code matching schemes, the result shows the algorithm hasobvious advantages on rotation invariance.