基于法医图像的阈值法分割的研究

通过对法医肾切片细胞的分割，统计计算肾脏细胞的面积，分析受害者肾脏细胞在显
微镜图像下面积随死亡时间变换的规律性，从而推测受害者遇害时间；基于直方图的阈值分割是一种有效的分割技术，其关键是确定阈值。分别采用了迭代阈值法、最小模糊集阈值、最大熵阈值、最小偏差阈值比较研究肾脏细胞的分割结果。实验研究表明，迭代阈值法相对于其他阈值分割算法处理法医肾细胞切片的分割效果较好，其原理简单，对图像的噪声具有一定的鲁棒性。
关键词：图像分割 迭代阈值法 法医图像 JAVA
Research of Legal Medical Image Segmentation with Threshold Algorithms
He Xiang-Qian (School of Basic Medical Sciences, Chong Qing Medical University, Chongqing, 400016) Abstract: The objective of this paper is to segment the kidney cell in legal medical image to get the area of each cell and then analyze the relationship between the average area of cells and the death time of casualty in microscope image. The Histogram-based threshold segmentation is an effective technology. The key parameter in the threshold segmentation is the choice of the threshold value. This paper develops the iterative threshold, the minimum fuzziness threshold, the maximum entropy, the minimum error threshold methods to compare the result of segmentation. The experiment shows the result of portioning the kidney cell legal medical image with the iterative threshold method is better than other threshold algorithms. The iterative threshold segmentation is relative simple and robust against image noise.
Key Words: Image Segmentation, Iterative Threshold Method, Legal Image, Java