Research on Classification and Retrieval of Digital Art Graphics Based on Hollow Convolution Neural Network

Digital art graphics are images created by people using computers through numerical calculation methods. They often have local and overall self-similarity, dynamic balance, symmetry and other characteristics, as well as fine structures, rich connotations, unique styles, and scientific and artistic double value. Based on the research object of digital art graphics based on atrous convolutional neural network, this paper extracts the aesthetic features of images by analyzing the relationship between image features and aesthetics of digital art graphics, and establishes an image feature library. Indexing algorithms classify and retrieve digital art graphics. Taking the current popular digital art graphics as the research object, the method of automatic classification and retrieval of image aesthetics using computer image processing technology is explored. Extract and fuse effective aesthetic image features; use SVM to establish image aesthetics classifier, and realize the classification of high and low aesthetics of images; use implicit semantic indexing method to achieve image emotion retrieval with different semantics; The effectiveness of the algorithm is verified by experiments.