自适应腐蚀模板在球形目标分割中的应用及实现

在图象处理领域中目标分割是目标识别过程中一项必不可少的工作。在某些情况下，待识别的目标与其它物体相连在一起并且灰度值相近，这样使用单纯的灰度阈值法很难将目标分离出来。对特定形状目标（球状目标）的分割技术进行了分析，提出了一种新的自适应腐蚀模板技术。这种技术弥补了由于模板大小固定而当目标的大小不同时，出现的模板相对过小不能有效地将目标分割出来，以及模板相对过大而将目标完全腐蚀掉的缺陷。详细叙述了自适应腐蚀模板的生成技术，解决了特定情况下的球形目标的分割问题，并应用Visual C++ 6.0进行编程实现。最后，给出了几种情况下的分割效果比较图。

Application and Realization of Adaptive Erodent Algorithm in Spheral Target Segmentation

In the field of image processing, target segmentation is necessary for target recognition. In some circumstances, the target to be recognized is connected with other targets and their gray level is similar. Thus, it's difficult to separate target using a single threshold. In this paper, segmentation method for characteristic target (sphere) was analyzed, and a new method for segmentation named adaptive structuring element was introduced. This method makes up for the drawback resulted from fixed structuring element for target with different size. The trait of open operation and the method of adaptive structuring element go into particulars, and the segmentation of sphere target is settled in special circumstances, and it is programmed with Visual C++ 6.0. At last, segmentation results were compared in several conditions.