基于自动选取最佳阈值的X光图像快速分割方法

New Quick Segmentation Method of X-ray Images Based on Finding Optimum Thresholds

为实现对数字X光图像进行快速分割,提出一种针对直方图的包络特征进行处理的新方法。该方法采用分水岭算法对数字X光图像的直方图一维信号进行处理。首先计算出图像的直方图,然后对直方图进行形态学滤波,提取出直方图的包络。在此基础上对直方图的包络采用分水岭算法自动提取出最佳阈值。用提取出的阈值进行基于阈值的初步分割,然后进一步采用基于标记的分水岭算法对初步分割的二值图像作进一步分割。通过实验表明,采用该方法可以有效地分割大幅X光图像。

A new method was proposed to segment the mammary X-ray images quickly. The method was based on the envelop curve of histogram. It dealed with the signal of histogram that was calculated by watershed algorithm. First, the histogram of image was calculated, and then the histogram was filtered by mathematical morphology to extract the envelop curve. At last, the thresholds were calculated by watershed algorithm according to the envelop curve. Then on the basis of the first segmentation using calculated threshold, X-ray image was segmented according to the marked watershed algorithm. Computer experiments indicate that this method was very effective to segment large scale mammary X-ray images.