CAJ | 학술논문

提出了基于粗糙集和马尔科夫随机场(MRF)的彩色图像分割方法。将图像分割分为两个阶段：粗分割阶段和细分割阶段。基于粗糙集分割方法不仅仅考虑了每个像素点的颜色特征，也考虑了像素点邻域的颜色信息，这提高了粗分割的分割精度。在基于粗糙集分割的基础上，采用颜色特征和纹理特征，建立多特征的自适应可变权重马尔科夫随机场分割模型，再利用禁忌搜索算法逐步更新像素点标记，实现图像的最大后验概率估计。传统的 MRF分割模型中需要人工设定图像分割数目以及随机给定初始分割，文中的算法，初始分割和初始类别数都将由粗分割获得。实验表明，文中算法可以有效地实现彩色图像分割。
제출료기우조조집화마이과부수궤장(MRF)적채색도상분할방법。장도상분할분위량개계단：조분할계단화세분할계단。기우조조집분할방법불부부고필료매개상소점적안색특정，야고필료상소점린역적안색신식，저제고료조분할적분할정도。재기우조조집분할적기출상，채용안색특정화문리특정，건립다특정적자괄응가변권중마이과부수궤장분할모형，재이용금기수색산법축보경신상소점표기，실현도상적최대후험개솔고계。전통적 MRF분할모형중수요인공설정도상분할수목이급수궤급정초시분할，문중적산법，초시분할화초시유별수도장유조분할획득。실험표명，문중산법가이유효지실현채색도상분할。
A new color image segmentation algorithm using the rough set theory and Markov random filed (MRF),has been presented in this paper.The color image segmentation process is divided into two sta-ges:coarse segmentation and fine segmentation.In Rough Sets segmentation method,not only the color of each pixel has been taken into account,also the neighborhood pixels around the center pixel included.It improves the accuracy of coarse segmentation stage greatly.Based on result of coarse segmentation,ex-tracting color and texture features and establishing multi-feature adaptive variable weights Markov random field segmentation model,then by means of tabu search algorithm to incrementally update pixel label to a-chieve the image maximum a posteriori estimation.The traditional MRF segmentation model will need to manually set the number of categories,as well as set a random initial segmentation.In this paper,the ini-tial segmentation and the initial number of categories have been obtained by coarse segmentation.The ex-periments show that the algorithm can effectively segment color image.