摘要
为提高对工业X射线图像的分割效果,提出一种改进的OTSU图像分割算法。分析了经典OTSU算法存在的不足,将类内方差也作为计算最佳阈值的一个因素考虑,从而提出了一种改进的OSTU阈值选择函数。在阈值分割过程中,采用局部递归法逐次进行分割。采用目标背景面积差和类内方差的比值作为递归结束的条件。通过与最大熵法、经典OT-SU法、局部递归OTSU法进行实验对比,证明本算法对于射线图像是一种优秀的阈值分割算法。
In order to improve the quality of segmenting the X-ray Image in the industry field,an improved OTSU is proposed.First of all,analyzing the shortcoming of the OTSU and proposing a new threshold function according to the suggestion that make the variance within clusters as a factor of computing the best threshold.And in the process of segmenting,the method of local recursion is used for recursive segmenting.The ratio between the difference value that acquired from the area of the objective and the background and the variance within clusters is used as the terminal condition of the recursion.According to the contrast with the KSW entropy,classical OTSU,the recursion OTSU,this method is very effectual for the segmenting of the X-ray images.
出处
《计算机工程与设计》
CSCD
北大核心
2012年第6期2388-2391,2397,共5页
Computer Engineering and Design
关键词
射线图像
阈值分割
类间方差
类内方差
递归
X-ray image
threshold segmentation
between clusters variance
variance within clustersrecursion
