Content-based Face Image Retrieval Technique

CBIR is becoming the hot research, and the Content-based face image retrieval technique, had some special means in both research and application field. This paper studies about in this subject. The main works of this paper is:Firstly, This paper filter out the interference of the light and the coulor of the skin by gray-scale transformation and binaryzation, then filter out the interference of the background by division. Then Strengthening the characteristics by normalization and linear interpolation. The interference of the image was filtered after those changes. After that extraction the feature of the face image. The result of the experiment that Gabor wavelet transform can extract the feature of face image in a effective way, was also show in this paper.Secondly, calculate the similarity of image, to search out the most similar image, then build the search system. It is a problem that the result of the search is not as good as we expect, so this paper used2kind of feedback method, and show the result of the experiment to prove that the modified feedback method was better than the traditional feedback method.At the end of this paper, we use a experiment to explain the strengths and weaknesses of relevant calculation, and finally get the best way to search the face image.