Color Image Segmentation Based On Semantic Color And Color Extraction Research

In image processing and computer vision, image segmentation is a critical technology, which is also an inevitable process of image understanding. In segmentation an image is expressed as a number of non-overlapping connected regions which provide necessary information for feature extraction, object recognition and other high-level image understanding. Because a color image provides more information than a gray image, color image processing is received wide attention. Meanwhile, as one of the most important features of image vision, Color is also an important way for human perception for things. And color is a key factor to reflect the semantic information of an image. Color recognition plays an important role in the image project, pattern recognition, computer vision, medical image.Based on human visual experience and semantic description of color, and the fuzziness of image itself, semantic color is research. According to the fuzziness of semantic description and color, the concept of semantic color is introduced. Through establishing the relationship between the characteristic vector of color and the semantics classification, realize the color and semantic description of mapping between and features expression, the mapping and feature expression between color and semantic description is realized. At the same time, semantic color model and semantic database is built, at last, color feature extraction and retrieval is achieved based on semantic color, combining with the fuzzy clustering algorithm of color image segmentation.In this article the significance and background of the research is firstly described, secondly the current research status about color image segmentation and image information retrieval based on semantic feature is analyzed emphatically according to the existing research results of color image segmentation. The method of color semantic description is be introduced according to human visual experience and semantic description of color, and the fuzziness of image itself, and the structural relation between color and semantic description is built. Then the semantic color structure and semantic database are established, which build link between low-level features and high-level semantic, thus proceed color image segmentation and query and retrieval of semantic color.