Semantic Image Retrieval Based On Contents And System Design And Implementation

Visual senses play an important role in the process of human obtaining information and80%human’s visual sensory information come from image.Nowadays, with the rapid development of computer science and technology, the image storage and retrieval dominate people’s daily life. however, the existing text-based image retrieval system can’t meet people’s requirement due to the backward technology by manual mapping. The content-based image retrieval technology, a newly emerging technology, however, has greatly enhanced the accuracy and speed of image retrieval by making up the deficiencies resulted from manual mapping, which is charactered by the abstraction of the highest similarity matches of image sequence. Nevertheless, neglect of people’s subjectivity when looking at the image which is of rich semantic meaning call sed by the content based image retrieval will result in the inaccuracy and incorrectness featured by "semantic gap". In recent years, many scholars have devoted to realize the people to machine intellectual dialogue and the seamless docking between customers and multi-media by narrowing the "semantic gap". One effective method is to add a recognition module to the vacant emotional sematic in the content-based image retrieval system, making the bottom layer image feature integrated with the high layer emotional sematics. Many scholars have made research and experiment in this new area by studying the content and emotional sematics image retrieval system and they have got many unique perspective and valuable experiment.however,the content and emotional semantic based image retrieval technology have to be systematized and more theoretical research and experimental data are needed when they are really put into practice.Based on the concept of "to make up which is vacant", the author of the thesis renders a innovative but simple viewpoint featured by adding emotional sematics recognition module in the content-based image retrieval system. focusing on the problem of "how to make up vacant", the thesis aims to find a way to reduce the "semantic gap" by compensating the retival deviation caused by the content-based image retrieval system.First, the author of the thesis has given a comprehensive analysis of principle of the content-based image retrieval system and the bottom layer feature abstraction skill, with the centers on the main algorithm of the abstraction of colour, texture and shape. the people is also well structured by managing to offer a framework for the content and emotional semantic based image retrieval system through the integration of the higher leval emotional semantic recognition module;and,by dint of the framework, the paper elaborates the establishment of the emotiom, the algorithm of emotiom, emotiomal module as well as emotiomal space. Furthermore, the establishment of the correspondence relationship of the bottom image feature and the higher emotional semantic have been put forward in a standard high leval to entrench the emotional semantic recognition mechanism. LFCM-SVM is found to be the best reflection method through the clarification of various reflection method of the bottom image featuresand the emotional semantic. Last, both a satisfactory result is acquired by the effective experiment and some experiment datas are obtained by depending on the Matlab platform, which has made a meaningful attempt to the reduction of "semantic gap".