Study On Classification Of Style Painting Based On Information Entropy

Art and painting has been a very important part of the development of human culture.It records the development and progress of human civilization,and plays a very important role in understanding and exploring history,art and culture.Today,a large number of art paintings have been spread along with the history of history,and it is the treasure of human civilization.With the progress of science and technology,many art paintings have been digitized,which has brought rich data for scholars to study style painting.However,faced with such a huge amount of data,to find the style of painting is kind of look for a needle in the ocean.Therefore,how to use the computer to effectively identify and classify the mass style painting is a very important topic.The traditional image classification and recognition algorithms are mainly based on the content of the image to match the similarity,and do not consider the characteristics of the artistic style of the painting.In this paper,a classification algorithm,based on information entropy,is proposed,which considers the overall style of the image.Information entropy is a concept used to measure the amount of information in information theory,and the information entropy of the image reflects the overall distribution of the gray value of the image.This algorithm is mainly based on the following three categories of image entropy to describe the style of the image as a whole:(1)a color entropy is proposed to describe the overall color characteristics of the image.The image is decomposed into Lab color space,a and B were extracted from the channel color values,calculate the entropy,and then set a weight according to the priority mobile number of color histogram,color entropy to generate the image color feature description.(2)A partitioned entropy is proposed to describe the feature of the overall spatial distribution of the image.The image segmentation by region into fixed size blocks,each block respectively to calculate information entropy,the information entropy of different blocks reflect the different regions of the image corresponding gray value aggregation degree,then the information entropy variance of all the blocks,which reflects the overall image gray value distribution of polymerization degree,which describes the distribution characteristics image space.(3)a contour entropy is proposed to describe the features of the edge contour of the image.The method is based on the Contourlet transform to decompose the image into different scales and directions of space,describes the image contour feature.The coefficients generated by the Contourlet transform correspond to the contour features of the image,and also have the characteristics of aggregated and discrete.Finally,Support Vector Machine(SVM)is used to train the entropy features of the image,and then the style painting is classified.The algorithm uses information entropy to describe the features of the image,which has the advantages of less feature dimension and faster operation speed,and ensures the accuracy of recognition.Because the image is scaled at any rate,it will not affect the overall distribution of the image pixels,so the algorithm also has the scale invariance.The selection of pyrography,comic,sketch,oil painting,ink painting,mural paintings and watercolor painting style of the seven classification,experimental results demonstrate the effectiveness of the painting classification algorithm of information entropy based style.