Study On The Technology Of Target Image Detection And Image Segmentation Based On High Dimensional Visual Feature Model

This paper studies the theory model and key technologies of target image detection and image segmentation technology based on high dimensional visual feature model. The proposed technologies of target image detection and image segmentation are applied in our information protection system of relevant engineering project, and the developing task is accomplished successfully.Firstly, this paper proposes the high dimensional visual feature model of target image according to the human visual system, and the detection theorem and its deduction are presented based on the model. According to theorem and its deduction, the problems of target image detection and image segmentation are transformed into multi-stage recognition of high dimensional features. The system of target image detection and image segmentation are established based on the theory model. The function equations of key modules in the system are discussed, and several implementation frameworks of the system are proposed. The following key technologies in the system are then deeply studied.Color image edge detection based on neighborhood change vector field: color edge detection plays a very important role on feature extraction of texture. Different from other edge detection technology based on gradient, a new edge detection algorithm using neighborhood change vector field is proposed in this paper. The model of image neighborhood change vector field is established by computing neighborhood changes on multi-direction and several computation methods of vector field are discussed. Based on the vector field, this algorithm uses the directional sharpness of neighboring change to describe color edge, and the fuzzy clustering on directional sharpness of pixels in the image is utilized to get the adaptive edge detection result.Fuzzy integrated measurement of high dimensional feature and fast genetic detection for image target: this algorithm points out that the target image region is the sub-region which make the integrated measurement function of color-texture feature to get the maximum in all candidate target regions extracted under the constraint of target shape feature. By integrating the fuzzy measurements of color and texture features, the...