Research Of Chinese Paintings Identification Method Based On Image Recognition

Chinese painting is the unique product of China’s history of civilization, is an important and unique way the Chinese people see the world and express themselves, with its unique artistic style, it has become the China’s culture and art treasures in the world. Traditional Chinese painting identification method is usually depends on the painting style of C hinese painting, calligraphy, calligraphy of relevant experts’ s subjective sensibility to distinguish authenticity, the most serious deficiency is much affected by their own emotional factors, lack of justice, the judgment of authenticity and quantitative identification reference. Identify the authenticity of Chinese painting based on image recognition algorithm can make the relevant experts have unity, justice, and quantifiable reference index and enhance the credibility of science appraisal of C hinese painting. C hinese painting is a unique form of art, compared with other artistic style, most painting style is subjective and abstract, the result of brush texture, color, layout, line such semantic features of C hinese painting can be the different. Again in this article, after the study of domestic and foreign research results then deeply studied the artistic style of image classification algorithms, adopting suitable algorithm and the experiment of C hinese painting style and art characteristics and distinguish authenticity identification, its thoughts and methods applied in the development and design in the traditional C hinese painting image identification system. First in the image pixel domain, extract features based on the overall artistic style characteristics of the traditional C hinese painting and obtain different histogram features, to compare the overall artistic style and local artistic style, the edge detection algorithm based on Sobel operator which can get more typical calligraphy features of the painter’s painting style.And get the global and local art style of Chinese painting, and the fusion algorithm based on information entropy, in order to determine the final identification result, choosing a artistic style characteristic of minimum entropy, eliminate factors such as content of C hinese paintings. Then using three layer wavelet decomposition for the traditional C hinese painting image to achieve painting art style and method of texture feature. Compared with other algorithms, this algorithm adopts the resolution of the different forms and at different situations, different image in transform domain of painting style has different frequency band structure, can obtain the depth information characteristics of traditional C hinese painting art style. So in different painters works manifest classified recognition difference style of painting, has better advantages. Finally based on the extraction of traditional Chinese painting art style characteristics, tectonic authentic work model of normal distribution parameters, the calculation to identify sample and established an original sample of "distance", to determine their similarity and identify its authenticity and give quantitative analysis of the identification of Chinese painting based on image recognition and provide quantifiable, more precise results. According to the characteristics of the oriental art paintings, such as calligraphy, layout, brush painting skills, puts forward the new image characteristics and provides a good theoretical basis for adequate and effective computer vision algorithm. Experiment proves that the feature extraction based on traditional C hinese painting art style using in the field of recognition of Chinese painting also has a certain robustness, can get the depth information of traditional C hinese painting art style and realized C hinese paintings identification theory, technology and method.