Research On Content-Based Image Retrieval

With the rapid development of the Internet technology and the prevalent consuming electronics, plenty of digital image is produced and released. Searching the source image quickly and efficiently in the multi-media database is a meaningful task. Many current industry image search engines (such as Google and Baidu),when they're searching information with the associated image letter to complete the search task with the image content, which brings the dissatified search results. Mean while,based on image retrieval techniques is expected to address the critical issue.Content-based image retrieval techniques is analyses the content of images, such as color, texture, and other important characteristics of the underlying vision, then establishes the indexed feature and stores them in the characteristics library. When this user only enquiries, their vague impression of the image description, they will found out large-capacity image database image.Based on the characteristics color image retrieval, the ways of Searching algorithm are Color Histogram, The Main Colors, Color Moments, Color Set, clustering, scan line projection, etc. algorithm, Color histogram is often used to retrieve one of the ways.The color of CBIR on Feature Extraction, Reash on the image of the shape, color, texture and other characteristics, color is the most notable, reliable, and stable visual features, image recognition is the main characteristics of Perception. Yet, relative to geometric features, the color image of the target neutron direction and the size of the changes are not sensitive, but with a very strong robustness. A description of color images is the most simple and effective feature. People are impressed by an image, usually in color images from the spatial distribution of the beginning.For the shape of the characteristics of image seach ways, from the edge detection and image segmentation, and the outline of the extraction areas such as the shape of the image retrieval as an important link has been studied.the key point is the color histogram the algorithm and improved color histogram algorithm, combined with the actual completion of the color histogram which is based on image retrieval system.