概率伪形态学的自适应参数设置研究

The Study on the Adaptive Parameter Setting of Probability Pseudo Morphology

概率伪形态学是一种新的形态学处理方法,与传统形态学的基本运算不同,它无需对原数据进行排列而求得极值,而是基于切比雪夫不等式估计邻域内数据的两个伪极值来定义腐蚀和膨胀结果,其中参数k可用来调节伪极值与实际极值的逼近程度,并使概率伪形态学方法兼具线性和非线性特性。通过对图像直方图进行规定化处理自适应地确定参数k,减小了直接在原直方图上估计参数造成的误差。与传统形态学方法对比,概率伪形态学方法的腐蚀膨胀结果较好地保持图像的形状和细节结构,同时在边缘提取中得到的结果较理想。

Probabilistic pseudo-morphology is a new morphological method differing from the traditional in the basic operational process:dilation and erosion.It does not need to rank the original data then obtain the extremes,instead,it estimates two pseudo-extremes of neighborhood data based on Chebyshev inequality to determine the operations.The parameter k can be used to adjust the approximation degree of the pseudo-extreme value and the actual extreme value,and makes the probabilistic pseudo-morphological method have both linear and nonlinear characteristics.By regularizing the image histogram can adaptively determine the parameter k,then reducing the error caused by estimating directly on the original histogram.Compared with the traditional morphological methods,the results of probabilistic pseudo-morphological erosion and dilation can keep better shape and structure of the details of the image,and the results in the edge extraction is also better.