Research On Image Retrieval Method Based On Color And Spatial Feature

With the development and popularity of digital and computer technology, digital images have become popular as a form of digital information and have been used for many aspects. For the mass of the image information, retrieval has become a very important means for access to useful information. This paper focuses on the image color and spatial feature , and then extracts and matches with these features to realize the retrieval.Color is one of the basic features of images, which is one of the most widely used image retrieval of visual features. Compared with other visual features, color feature has less dependence on scale, direction and view port. The extraction of color features is relatively simple, has become a powerful image analysis tool. This paper achieves the interaction interface of the image retrieval system based on the color feature and tests the system with two sets of image library. Experimental retrieval based on user submitted sample to return to their same or similar color feature of the search results, and analyze the retrieval performance with rate of recall, precision and so on.The spatial relationship between the objects in the image is also an important feature. Especially in some images that distinguish between different details, the extraction and match with the spatial position features of the small objects becomes particularly important. This paper uses a group of label images as the research object. The method uses the partial iteration threshold to get the binary image in order to overcome the interference from light, background and so on. And then, the mark points are recognized. This method uses detail objects as the goal of feature extraction as the basis to retrieval and match, realize the detection and distinction. On the right and wrong circumstances, has made the detailed analysis.