基于区域生长法的医学图像分割研究

医学图像存在感兴趣区和背景区，感兴趣区是分割的重点。针对基于区域的分水岭分割算法中通常存在的严重过分割现象，而采用2点措施来改进：一是以梯度图像的浮点活动图像取代梯度图像进行分水岭变换，使边缘定位更准确；其二是以基于面积和对比度控制的合并小区域准则，有效抑制过分割现象，同时满足感兴趣区的分割要求。实验表明该算法简单有效，能够得到符合人类视觉系统特性的分割结果，边缘定位精确度较高。

Medical image segmentation based on region growing arithmetic

This paper presents a detailed analysis of still medical image segmentation. The medical image has two kinds of areas:the interesting regions and the background, and the interesting regions is the key point of segmentation. To deal with the extreme over-segmentation problem of region-based watershed methods, two approaches have been proposed. Firstly, we perform the watershed transform on the floating-point moving image, but not the morphological grad image, which can be located at the contour more accurately. Secondly, we propose a region merging rules based on the area and contrast, which can restrain the over-segmentation problem effectively and reserve the interesting regions. The experimental results demonstrate the simplicity and effectivity of the approach, and the results are in accordance with our understanding on the semantic concepts and the perceptual mechanism of human eyes.