Shape Analysis Of The Bone Inscription Based On Zernike Moments

The bone inscription is another important archaeological discovery after Anyang Oracle,1000years earlier than Oracle, which forwards the emergence of the ancient Chinese characters a thousand years. The history of Chinese characters was also rewritten probably. Making use of computer technology can help quantitatively analyze the shape of the bone description. It provides important support for further interpretation and study of the ancient writing.The paper first introduces the background and significance of the research, discusses the research status of image classification and retrieval technology. Then, a detailed description of using improved Zernike moments to describe bone inscriptions image is described. The support vector machine (SVM) method is used to classify and analysis the bone inscriptions image set. Finally we discuss the relationship between bone inscriptions and Yi by using the feature information.The main research content and contributions of the thesis are:Firstly, we introduce several feature description methods, including color feature, texture feature, shape feature and spatial relationship feature. The description of the Zernike moments is analyzed based on its invariance. An improvement is made according to the insufficiency of traditional Zernike moment in the affine invariants. The experiments show that the improved method keeps the original scale, translation and rotation invariance and at the same time, provides good affine invariance. This can enhances the accuracy of shape classification and retrieval.Secondly, according to the characteristics of SVM for small sample size, simple structure, global optimal solution, applicability, we adopted SVM classification algorithm. A sample set is used for automatic learning and a structure classifier training model is build. Then the classifier model is used to classify the test image in bone inscriptions image library.Finally, some commonly used similarity measure methods are discussed. According to the results of the binary flag image retrieval experiments, we use the improved Zernike moments features and weighted Euclidean distance metric to compare and analyze bone inscriptions image and Yi. It gives objective estimation to the inherit relationship between the bone inscriptions and the Yi.