基于区域生长的生物图像分割方法

Biological image segmentation method based on region growing arithmetic

针对基于区域的分水岭算法中通常存在的严重过分割现象,采用两点措施来改进:其一是在分水岭分割之前采用迭代阈值分割算法对图像进行分割,从而有效地抑制了由图片表面的灰度变化引起的过分割,使边缘定位更加准确;其二是在分水岭分割之后以基于面积,对比度和深度控制的区域合并准则,有效抑制过分割现象,同时满足感兴趣区的分割要求,实验表明该方法简单有效,能够得到符合人类视觉系统特性的分割结果。

To deal with the extreme over-segmentation problem of region-based watershed methods, two approaches have been proposed. Firstly, we use iterative threshold segmentation treat to image before watershed algorithm, thus effective inhibition over -segmentation caused by grey level change of the surface of picture, which can be located at the contour more accurately. Secondly, we propose a region merging rules based on the area- contrast and depth, which can restrain the over-segmentation problem effectively and reserve the interesting regions. The experimental results demonstrate the simplicity and effectively of the approach, and the result are in accordance with our understanding on the semantic concepts and the perceptual mechanism of human eyes.