Research Of Chinese Calligraphy Identification Method Based On Image Recognition

The research of simulating Chinese calligraphy is not only significant for the inheritance and development of the Chinese traditional culture, but also helpful for Progress of computer art and popularization of calligraphy art. As calligraphy works’ verification has long been investigated by calligraphists subjectively from the art’s field. How to distinguish the genuine calligraphy works from those of forged with objective measures and evidences?This paper starts with the original image of the scanned compositions, with the processing sequence of preconditioning, morphological processing, style characteristic quantization, style characteristic probability diagnoses to deal with the picture successively and finally give the authentic calligraphic works of probability.According to the difficulty of calligraphy extraction in stroke overlapping part and ineffective extraction of calligraphy stroke primitives, this paper puts forward a stroke extraction algorithm based on fuzzy region detection and improves the effectiveness of calligraphy stroke primitives extraction. Firstly, it uses the Candy operator to get the calligraphy contour and uses the skeleton of calligraphy to determine the approximate center of fuzzy region in the skeleton characteristic point. Secondly, it does the fuzzy region detection on the contour information around the approximate center point as the center. After detecting effective point in the contour, it uses stroke segment recovery rules to merge effective stroke primitives.Calligrapher through years of practice, in the creative writing will have their own style,and these characteristics are stable. For this feature, according to the obtained calligraphy stroke primitives, a series of rules from stroke level and structure level of the calligrapher can extract a number of feature points to characterize the calligrapher’s style of writing. Then, for each feature point using the Gauss distribution model, for feature point of different calligraphers using selection and weight estimation, structures character style feature vector, establishes different calligraphers’ style characteristic model. When calligraphy is distinguished, Calculated for the authenticity of the traditional method ofidentifying complex, computationally intensive shortcomings, the authenticity of identification algorithm based on the normal distribution, the probability calculation is converted to distance operations. On the basis of characteristics of the calligrapher style,authentic models constructed using the normal parameters, calculated to be the "distance" to identify authentic works and has been determined to identify the authenticity of the model.The experimental results show that this method can effectively extract the feature vector for each calligrapher’s style is given until the authentication status of a work.