基于“灰度-梯度共生矩阵”的最大条件熵阈值法  

Maximum Conditional Entropy Threshold Algorithm Based on Gray-Gradient Co-ocurrence Matrix

作　　者：张弘[1] 范九伦[1] 

机构地区：[1]西安邮电学院自动化学院,陕西西安710061

出　　处：《现代电子技术》2010年第20期49-53,56,共6页Modern Electronics Technique

基　　金：国家自然科学基金资助项目(60572133);陕西省自然科学基金资助项目(2007F18)

摘　　要：基于"灰度-梯度共生矩阵"模型,在现有最大条件熵图像阈值法的基础,引入加权系数进行改进。为了解决权值选取问题,以图像分割质量评价的均匀性测度为评价指标,采用自适应粒子群算法对权系数进行优化选择,进而获得最优的分割阈值。实验结果表明,与二维最大熵、最大条件熵算法相比,该方法能够获得更佳的分割结果。According to the maximum conditional entropy thresholding method, a weighted factor is proposed based on the ＂gray-gradient co-occurrence matrix＂ model. For the selection of the factor, a homogeneity measure is used as the image seg- mentation quality assessment, and an adaptive particle swarm optimization algorithm is used to select the weight coefficients to obtain the best segmentation threshold. The results show that the method mentioned above can get the best segmentation re sults in comparison with the the 2-D maximum entropy method and the maximum conditional entropy algorithm.

关 键 词：图像分割 灰度-梯度共生矩阵 条件熵阈值法 粒子群优化 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]