Research On Authenticity Identification Of Chinese Painting Based On Image Recognition

Chinese painting as a unique art form in the world’s art treasures, is also an important part of Chinese traditional culture and art. With China gradually draws into the international auction market in recent years, there has been a lot of fakes. So, how to accurately determine the authenticity of a Chinese painting has become a major problem of the development of Chinese painting at present. Traditional Chinese painting is mainly identify the authenticity by experts through its own existing experience, based on the works of calligraphy, as well as Mexican law which years of paintings by perceptual judgment, however, this method is vulnerable to subjective factors interference and lack of objective identification of quantifiable indicators. To solve these problems, the paper using an image processing computer vision technology, combined with some of the quantifiable metrics to identify the authenticity of Chinese painting, and designs Chinese painting authenticity of identification method based on image recognition, to assist in identifying the authenticity of Chinese painting work to improve the authenticity identification of credibility. The main work is divided into the following two aspects.(1)The paper designs the method of ROI(Region Of Interesting, abbreviated ROI) extraction according unique painting techniques and rich style type of Chinese painting, The new method analysis the characteristics of each ROI features extracted by low-level feature, then uses optimal weights algorithm to extract ROI feature image, ultimately, adjust the weight according to the contribution value of each feature in the image, optimal weights corresponding ROI is characterized as a style of the image.(2) The paper designs the authenticity of identification method based on the style characteristics of the region discrimination on the Chinese painting, establishes regionallevel Markov model due to MRF showing the relationship between the current value and its neighborhood. Wherein the first layer of the model uses an improved iterative belief propagation algorithm to get region authenticity state, the second layer, gets the authenticity of the state of the object based on the authenticity of the block status area, and in the third layer the status of the entire work is determined by making vote ultimately.Finally, under the Windows 7 operating system environment, programming to verify the feasibility and applicability under VS2012.