The Reaserch Of Image Retrieval Techniques By Color And Shape Based On MPEG-7

The image resources become more and more abundant in modern information era. The traditional text-based retrieval method can no longer meet the needs of image retrieval. Content-based image retrieval (CBIR) technology emerged as the times require, which combines many fields of knowledge, such as image recognition, computer vision, database technology and artificial intelligence. The key issue of CBIR is extracting features from low-layer image features such as color, texture, and shape to analyze the image content. MPEG-7 is a standard for describing multimedia content, which describes image with its content, and enables effective, efficient query and access to multimedia content.This paper focuses on the bottom basic features offered by MPEG-7, and the goal is to develop a system integrated the shape and color descriptors, which can be applied to content-based image retrieval rapidly, effectively. The main job is as follows: First, for the image segmentation is a key step in shape feature extraction, a new method of image segmentation was proposed based on the traditional threshold in this paper, and the feature extraction effect was improved by the method. Secondly, two methods were proposed for the sake of combining color and shape features to retrieve images, and the results of which combining two descriptors are superior to that obtained from single feature. Third, In order to improve the retrieval accuracy, the relevance feedback module was imported to the image retrieval system. Fourthly, integrated the color and shape descriptors offered by MPEG-7, as well as the relevance feedback technology, a practical and excellent content based image retrieval system was designed with VC++ and database technology.