Research On The Key Technology Of Digital Art Graphics Classification And Retrieval

Digital Art Graphics are image created by computers,using numerical calculation method,they have part and overall self-similarity,dynamic balance,symmetry and meticulous structure.These graphics have rich connotation,unique style and contain value of science and art.Digital Art Graphics naturally have complex features such as color,morphology,and texture,they are as abstract as traditional works of art,and have been a modern way to express nature.When we design or create,we hope to find graphics meeting the ideas and creative inspiration in the image database,but most of the patterns cannot automatically meet the users' subjective thoughts and feelings.Therefore,we need to use digital image processing technology,use computer to aesthetic classify Digital Art Graphics objectively and realize emotional retrieval.Not only can facilitate the designers to select the appropriate patterns,but also plays a guiding role in the design of pattern,inspire inspiration.Through the research of aesthetic characteristics,we explore new feature computing method that uses computer recognition technology for automatic image quality classification and semantic-based image retrieval.In this article,we choose the Digital Art Graphic as the theme,analyze and research the relationship between the features and aesthetics,use emotional model of psychology as reference to extract aesthetic features of images.We use the SVM method to achieve image classification of high and low aesthetic,apply semantic index method to achieve image retrieval.The work done is summarized as following aspects:(1)Focus on psychology research,images have been subjectively scored by multiple testers,then establish libraries by human aesthetic information,including two image databases-aesthetic level database and emotional semantic database.(2)The paper research on the Digital Art Graphics.In view of the inherent nature and characteristics,we extract image characteristics from high-level aesthetic,low-level aesthetics and statistical.(3)Establish the evaluation model of aesthetic characteristics of images,the SVM method is used to build aesthetic classifier for the extracted features.(4)Establish image emotion characteristic model,introduce LSI implicit semantic indexing algorithm,to achieve emotional content retrieval of images.In this paper,we choose Digital Art Graphics which are popular nowadays as research objects,explore new feature computing method that uses computer recognition technology for automatic image classification and retrieval.Extract and fuse affective aesthetic visual features,use the machine learning method to build image aesthetical classifier,achieve image classification of high and low aesthetic,apply LSI semantic index method to match the similarity,verify the validity of the algorithm.The research of this paper can be applied to related research and development of practical evaluation and retrieval system.