基于云的过渡区提取与图像分割算法

提出了一种基于云的图像过渡区提取算法.该方法利用区域生长算法提取目标内部灰度同质性及相似性最高的区域作为云核,结合云发生器,对图像进行云化处理.通过相邻云之间的逻辑运算生成边界云,提取边界云的数字特征,利用边界云的熵来确定过渡区的两端极值.在此基础上,采用过渡区象素的直方图峰值对应灰度作为分割阈值并进行图像划分.试验结果表明,该算法在能取得较好分割效果.

Cloud-Based Transition Region Extraction and Image Segmentation Algorithm

A transition region extraction algorithm based on clouds is proposed. First, the regions with maximum gray homogeneity or similarity are extracted with the region growing algorithm as the cloud nucleus. Second, a cloud model is constructed for each object in image by a cloud generator. Third, boundary-clouds are constructed by Boolean calculation between two or more clouds, their digital characters are obtained and the two extreme values of the transition region are decided by the entropy of the boundary clouds. Finally, the image is segmented by the histogram peak gray value of the transition region. This algorithm, when tested in experiments, has proved both efficient and effective.