机构地区: 安阳师范学院计算机与信息工程学院
出 处: 《数据采集与处理》 2010年第2期165-170,共6页
摘 要: 针对常见的多分辨率分割算法在每一尺度分割过程中信息利用不充分的问题,提出了一种基于小波分解的变尺度多分辨率纹理图像分割新算法。该算法在每一尺度的分割过程中充分利用了各尺度上的有关信息:通过变尺度特征场考虑了更高分辨率尺度上的特征数据;通过变尺度标记场考虑了更低分辨率尺度上的标记数据。从最低分辨率尺度到原始分辨率尺度逐次进行图像分割,低分辨率尺度的分割结果通过直接投影作为相邻的更高分辨率尺度的初始分割,最高分辨率尺度上的分割结果作为本文方法的分割结果。实验表明,该算法具有较好的分类性能。 To overcome the insufficient information exploitation of the classical multi-resolution segmentation method, a new algorithm is proposed on variable scale based on the wavelet decomposition. In the algorithm, the segmentation on each scale can make full use of information on all scales; the feature data on the finer scales is considered by the scalable feature field; and the label data on the coarser scales is integrated by the scalable label field. The image segmentation procedure is sequentially executed from the coarsest resolution scale to the finest resolution scale, and segmentation result of the coarser scale is directly projected on the nearest finer resolution scale as its initial segmentation. The segmentation result on the finest resolution is used as the final result of the algorithm. Experimental results show its better performance on the image segmentation.
领 域: [自动化与计算机技术] [自动化与计算机技术]