A Survey On Content-Based Image Retrieval Of Intelligent Techniques

With the development of computer technologies and the advent of the World-Wide Web, there has been an explosion in the amount and complexity of digital data. It's very important to retrieve such multimedia information as digital images, video, audio, graphics, and text data efficiently and effectively. In image retrieving, the computer can get the meaning of the image only after it has processed, analyzed and explained the data. It will be the main research work of the computer vision and image comprehension, which has not been accomplished till now. In view of this problem, ways of content-based image retrieval (CBIR) has been found out, which mainly make use of the elementary audio features of image itself, such as color, texture and shape features, etc. to express the information included in the images and to complete the images retrieval.In this article, I tend to give a brief introduction and analysis to the principle, system and technology advancement of the CBIR. I think that the practical and intelligent retrieval method is to recover the features parameter which can express the contents of images. Besides, I have calculated the similarity metric and automatically classify the corresponding video images in the image database and feature base. CBIR is mainly made up of features extraction module, data base establishing module, user inquiring interface and feature matching module, whose main techniques are human computer interaction technique, image features description, retrieval and matching technique, data base organization and management technique, retrieval evaluating technique, etc.CBIR, which includes the extraction and analysis of image features, image edge recognition and detection, etc., has investigated such problems as how to establish image index, calculate and inquire similarly distance of images and target images, index according to similarity matching, etc. On the other hand, in order to make the comprehension and recognition of computers to images more accurate, we need to collect advanced semantics from the gained lower video features. Now, the common ways are content-based image classification, clusting and relevant feedback based on human computer interaction, etc.Finally, using rough set theory, the paper suggests an applying idea and methodin the relevant image processing. The main idea of rough set theory is to classify knowledge by indiscernibility relation, to describe concept with a lower and an upper approximation and to find out the decision and classification principles by knowledge reduct. As to the collecting of the intelligent images, its main idea is to choose features of images with the idea of attribute reduct and to enhance and reduce images using the equivalence relation.