The MRF Image Segmentation Algorithm Based On RGB Color Distribution Model

The purpose of using image processing technology is to modify the graphics,improve the quality and extract effective information of the images,image segmentation is one of the most fundamental and important areas of image processing.At present,the technology of gray image segmentation with using Markov random field model has been well developed,because color image carry more information,most of the existing segmentation methods have the problems of slow segmentation speed,inaccurate segmentation results and low universality.In order to solve this problem,this paper combines the MRF model,uses RGB color distribution model to describe color information instead of the traditional gray level model,and some research on image segmentation have been done.(1)Comparison and selection of segmentation models.Several image segmentation algorithms which are commonly used are introduced.After comparing their advantages and disadvantages,the MRF random field model which with fewer parameters,good noise immunity,stable model and easy combination with other models was chosen.The progress of MRF theory which is combined with other models in the field of image segmentation is studied,and the advantages and disadvantages of these algorithm are analyzed.Based on the idea of combining MRF theory with other models,it is applied to color image segmentation.In order to obtain a better effect of color image segmentation,this paper proposes a color image segmentation algorithm based on MRF theory.(2)Modeling and parameter selection of image segmentation model.This paper introduced the MRF basic theory,including the definition of the neighborhood system,the group in the model,and the method of parameter estimation.Then studies several common MRF models,and the optimal segmentation standards of images were introduced,and the advantages and disadvantages between them were compared.Several commonly used color space models are compared,and their advantages and disadvantages are compared,In order to select a suitable one to describe the information of color image pixels.RGB color model is chosen as the color space model of the algorithm,after considering the actual situation of the research subjects.(3)Optimization algorithm.In this paper,the planar MRF model and the hierarchical MRF model was combined with the RGB color space.The method presented in this paper expresses the spatial relationship between pixels by using the principle of MRF,and fully expresses the information of pixel values in the RGB color space,which has improving the accuracy,adaptability and rapidity of the segmentation.To prove the superiority of the algorithm in this paper,the FCM algorithm is compared with the proposed algorithm under the same operating environment and segmentation target.It is proved that the proposed algorithm is superior and effective in computing speed and segmentation precision.