摘要
利用分层马尔可夫随机场模型表征纹理图像，根据最大后验概率分布准则提出一种确定型松弛分割算法。对纹理图像进行分割的实验表明，该算法与模拟退火算法相比，明显地减少了计算时间，其分割误差与模拟退火分割误差非常接近。
A hierarchical Markov random field model is adopted to characterize textures. A deterministic relaxation algorithm based on the maximum a posteriori(MAP)criterion is presented for texture segmentation. Some experimental results show that the algorithm can reduce the computational cost significantly, and that the errors of segmentation using it are close to those using the simulated annealing.
出处
《北京师范大学学报（自然科学版）》
CAS
CSCD
1994年第1期60-66,共7页
Journal of Beijing Normal University(Natural Science)
基金
国家自然科学基金
关键词
马尔可夫随机
纹理分割
图象分析
Markov random field model
Gibbs distribution
maximum a posteriori criterion
texture segmentation
deterministic relaxation
simulated annealing
