Research On Color Based Image Retrieval

With the continuous development of digital technology, multimedia digital image both in quantity and scale are in rapid growth, how to manage these massive amounts of image resources has become an important problem faced by people. In terms of image retrieval technology, the traditional text-based image retrieval technology already cannot satisfy the needs of many practical applications. In recent years, the technology of content based image retrieval has become a new research hotspot.Color is one of the main characteristics of human eye perception, the statistical features of color is also a kind of effective image representation, as a result, the method of image retrieval based on color has became one of the most representative retrieval technologies. Based image retrieval as the goal, the quantification of HSV color space, color and monochrome interval division of related research, such issues are the main research contents and research results are as follows:1. Present a new color quantization algorithm. Taking the HSV color space as a whole and according to the characteristics of the human eye color, color space is divided into10interval, using Microsoft’s216safe color as the color chart color characteristics of image statistics, which can bring out an increase in feature dimension and reduce the computational complexity in the process of feature extraction and matching color. The experiment shows a better search result.2. Give a new colorful and gray region partition method. By using image processing software, color saturation and brightness change impact analysis, color area division and grayscale curve is given. The experimental results show that the retrieval results of using the new division method is good, the rate of recall and weight significantly is higher than the traditional classification method.3. Implement an image retrieval prototype system based on color feature. It takes OpenCV1.0and Visual C++6.0as the system software development tools, which has a good interface and a strong practical function. The interface concludes the image display window, the system function button and the display search results window, etc. Retrieval module embedded in a variety of different ways we can compare different retrieval function of them。...