Simulation And Analysis Of Texture

In this paper, we first introduced the concept, characteristic and development of thetexture, then some general methods of texture analysis are presented, such as statisticalmethod. Then various methods for simulating texture such as CM, SAR, MA, LC andGibbs method are analyzed and discussed, with emphases on LC method and Gibbsmethod, and the advantages and disadvantages of the two methods are compared. Gibbsmodel is a more flexible method and can describe the image information morethoroughly over the above methods. The interaction structure and potential can beobtained by recovering the characteristic clique family instead of being defined ahead oftime, then the structure is applied to set up a model, the thorough information of eachkind of texture can be obtained. In the application of Gibbs model, the CSA method ischosen to adjust the gray-scale of pixels, which avoids excessive parameter estimationand can obtain the variation of gray-scale of pixel directly. At the same time theMetopolis method is used to realize the random relaxation of pixels, then somesimulation results are presented, and on the basis of which the analysis of first-orderstatistics, second-order statistics as well as difference statistics is made. It can beconcluded from the comparison that the simulated images are in good agreement withthe original ones.