Image Retrieval And Applications Based On Color And Texture Feature

Content-based image retrieval is the current active research, has applications in many fields. Specific content based retrieval process is according to provide the sample image or draft, the computer automatically find the characteristics of the sample figure, and the similar images will be return back. So based on image content retrieval is one of the key technologies of image feature extraction, feature extraction is based on the content based image retrieval, feature selection is good or bad for the image retrieval system has an important impact, so you need to choose a good feature extraction method can effectively to image retrieval. In recent years there have been many scholars about feature extraction was put forward many methods, such as color features, texture features, shape features, etc, and the practical application in every field has played a very good effect.In this paper, on the basis of past research, according to color features and texture features in-depth research and experimentation. For color features the main use of method is color histogram for image retrieval, summed up some problems by experiment which exists in the process of color histogram in the image retrieval. Then using gray level co-occurrence matrix of image texture feature and LBP algorithm to image retrieval experiments, concluded the use of the advantages and disadvantages from the experimental gray level co-occurrence matrix and LBP method. We can see it from the experiments of only using an image retrieval method to retrieve image retrieval, the effect is always not ideal, so this thesis combines the color image retrieval and the texture image retrieval, which is comprehensive using in image retrieval, comparing the effect of the three kinds of image retrieval, the experiment summary can be seen from the chart of using two kinds of retrieval methods can make the image retrieval accuracy, and the image of the retrieval results more close to the original image, comprehensive using two kinds of retrieval methods can overcome the existing disadvantages based on the original which is an image content retrieval method, and the retrieval results more close to human visual perception. So this integrated approach can be widely used in medical research, when examine patients, the image will be more clearly and test results is more accurately. Finally the simple introduction to shape feature of image content will be introduce in the paper.