Image Segmenting Technology Research And Application Of Color

Image segmentation is the main problem of image processing, and also in the field of machine vision. The color image is the most realistic description of the external objective world. Until now, there has been a considerable number of results and conclusions for the gray-scale image segmentation, but the study of the color image segmentation is relatively less due to it is relatively complicated.The image is divided into several groups(called region) that are non-overlapping, internally connected by the image segmentation according to the similar and uniform criteria. Similar to this, the color image is differentiated into one or more groups that are internally connected and uniform(homogeneous), and the uniformity of criteria is based on the features extracted from the image spectral components. The components are defined in a given color space. The process of segmentation is built on the knowledge of the identification of the target in the scene, such as the geometric and optical characteristics.The difference between the gray-scale image and the color image from the segmentation method is mainly manifested in the description of each pixel, the pixel in the gray-scale image is described by the one-dimensional brightness space, and the pixel in the color image is described by the three-dimensional space. The color image contains more information than the gray-scale image, therefore, the color image segmentation is very necessary in many cases.The major research work in this paper are as follows:1.Discussion on the shape of white areaThe color measurement and statistics, which include RGB color coordinate system, XYZ color coordinate system, CIE chromaticity diagram, JND and uniform color system, was introduce briefly. By fitting the white area boundary of XYZ chromaticity diagram and describing the shape of white area boundary in the uniform color system, this paper could provide a basis for the general color worker, and demonstrate the limitations when judging the white object with chroma difference and color difference.2.The feather color grading based on support vector machine In order to eliminate the shortcomings in feather color grading, such as the large error, a novel intelligent grading method is presented, which is based on the support vector machine(SVM) algorithm and the neural computation technology. The main process is to acquire the characteristic values of the feather’s surface as the feather’s color features by the computer vision technology; based on the SVM theory, the RBF function is selected as the kernel function to design the grader; finally grade the feather with the grader by inputting the feather’s color features.The method were implemented by VC++6.0software, and a large number of feathers were graded in the industrial site, the results showed that the method met the industrial feather color grading very well.3.Segmentation of feather stainThis paper segmented the feather stain by using the color difference method in the LAB color space and least square method under the gray image, and compared the effect of them. The effect demonstrated that the color image contained more information than the gray image, and the color image was the most vivid description to the objective world.In this paper, the above-mentioned algorithms and methods were implemented by MATLAB and VC++6.0software, finally, the experiments of the color image segmentation were verified, and the results were good.