Realization Of Face Image Segmentation Based On Image Retrieval

With the rapid development of computer and network technology, images, video, audio and other multimedia information technology are changing with each passing day, people can follow their inclinations of the use of a variety of network and multimedia resources to improve their colorful life. The progress of information technology in today’s society has exceeded people’s imagination, how to use a variety of image resources to further extract information they are interested in has become the focus of attention, in order to be able to accurate and convenient retrieval and access to image resources, image retrieval technology was born. This paper studies the human face images segmentation based on image retrieval technology, using face segmentation technology to make the complex facial image regionalization, then carries on the characteristic analysis, through the analysis of characteristic value, and selects the suitable for realization of retrieval technology character description.In this thesis, we use a variety of methods to extract facial features, because of the complexity of the facial feature information, a method of feature extraction cannot satisfy the requirement of accuracy of feature information, so this paper adopted the face skin color feature extraction method, the facial geometric feature extraction method and the face texture feature extraction method, the accuracy of the face feature is greatly enhanced, we choose the extract characteristic information for image retrieval of data to provide data support for the next step of image retrieval. Among them we use the method of color characteristics of face identification, to segment the whole face, which was carried out on the face split again, to get the corresponding characteristic information. Face image preprocessing needed to be done in the first place, respectively the color feature extraction of face features, geometrical feature extraction and texture feature extraction, getting the corresponding characteristic information, and according to the characteristics of the corresponding information to face image segmentation, in order to test the accuracy of the data obtained.In the end of the paper we choose different face images for feature extraction, and analyze the feature with noise in face image, analyze and establish the united face feature extraction system, combine the three kinds of feature extraction method efficiently and achieve the feature extraction accuracy requirements.