The Research Of Image Retrieval Based On Edge And Color Feature

With the development of Internet and multimedia technology, Especially in the growing popularity of digital cameras, scanners and other multimedia devices, the number of digital images increase rapidly, there is an urgent need to solve problems that how to quickly and effectively to retrieve the images which the user needs from a massive image database. Therefore, content-based image retrieval techniques have emerged, and it become hot in the field of image retrieval research.Content-based image retrieval is the technology that use the image color, texture, shape and other visual features, It can effectively make up for the defect of the text-based retrieval technology, directly on the image content to analyze image and extract the image feature, and then use these features as an index, in the image library according with a certain degree of similarity measurement method to search. Image retrieval algorithms is the core of content-based image retrieval.This article mainly focuses on image retrieval algorithm, and research the method of image retrieval based on edge and color feature. The main work of thesis including the following aspects:The first, described the technology of content-based image retrieval and the status of development in detail, mainly studies the extract method of color, texture, shape, position and other features, the measure of similarity, the evaluation of CBIR and relevance feedback.The second, study and analyze the method of the classical image edge detection in detail, Point out the merit and the scope of application of Roberts operator, Sobel operator, Prewitt operator, the direction operator, Laplace operator and LOG operator. Also studies the Canny operator and its defect, these methods are analyzed carefully trough experiment simulation, and also analyze every operator's effect in the noise.The third, in the image retrieval algorithm based on Canny edge detection, the quality of edge directly determine the image retrieval, studies the improved methods of Canny operator in the edge detection, through experiment and test result, it demonstrates the method is an efficient approach. Then propose a method of image retrieval based on Canny adaptive edge extraction and color feature, this method use Canny adaptive operator to extract the image edge to search the database, find similarity with the sample image and form a new image database, then used dominant color to divide image and decide target block in the new database, used the target block to detect the new database. It demonstrate that this method can improve the image retrieval performance, and also verified the feasibility and effectiveness of the algorithm.