摘要
在小波域隐马尔可夫树模型(WHMT)图像分割算法的基础上,根据二代bandelets系数的分布统计特性,提出一种新的bandelets域隐马树模型(BHMT)有监督图像分割算法。新算法试图利用二代bandelets能表示图像几何规律性的特点来改进分割效果。为证明算法的有效性,用合成纹理图像和真实的航空和遥感图像做分割实验,并和小波域隐马模型树分割算法作了对比,对合成纹理图像分割,用错分概率作为评价参数。实验结果表明,此方法错分概率低、分割效果理想。
Based on wavelet-domain hidden markov tree model and the characters of second generation bandelet coefficients, a new supervised bandelets-domain hidden markov tree model algorithm for image segmentation was presented. The algorithm was intended to improve the performance of segmentation by the use of the characters of bandelets with geometric orthogonal filters. In experiments,Synthetic mosaic image,aerial image and SAR image were selected to evaluate the performance of the method, and the segmentation results were compared with wavelet domain HMTseg method For synthetic mosaic texture image,miss-classed probability was given as the evaluation of segmentation results. Experiment results show that the method has lower missed classed probability and better performance.
出处
《计算机科学》
CSCD
北大核心
2009年第1期218-221,共4页
Computer Science
基金
国家自然科学基金(No.60372050)资助
关键词
图像分割
带状波
隐马尔可夫树模型
多尺度
Image segmentation, Bandelets, Hidden Markov tree model, Multiscale
