摘要
为实现CR图像实时快速分割,提出一种针对直方图的包络特征进行处理的新方法。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改进的基于标记的分水岭算法自动提取出最佳阈值。最后根据从直方图中提取出的、代表图像特征的阈值对CR图像进行实时快速分割,在采集完图像的同时就可以得出分割结果。通过实验和时间复杂度与空间复杂度分析表明,采用提出的算法通过减少运算量,可以快速分割出CR图像并提取出感兴趣区。
A new method was proposed to segment CR images quickly.It deals with the envelop curve of histogram. First, the histogram of image was calculated, and then the histogram was filtered by mathematical morphology to extract the envelop curve. At last,the optimum threshold was calcuated by watershed algorithm according to the envdop curve and then CR image was segmented quickly by calculated threshold that presents the feature of image. Experiments and the analysis of complex in time and space proved that this method is effective and quick to segment large scale CR images. Compared with other methods, the method can reduce the burden of computation, so it can segment CR images quickly.
出处
《光学技术》
EI
CAS
CSCD
北大核心
2007年第1期6-10,共5页
Optical Technique
基金
国家自然科学基金项目(69677015)
关键词
CR图像
直方图
最佳阈值
分水岭算法
CR image
histogram
optimum threshold
watershed algorithm
