基于小波包分解的白细胞胞核边缘提取算法与实现

白细胞胞核的形状是病理学家观察和检查白细胞样本的重要依据,是白细胞自动分类技术的关键.因为通过它可以区分白细胞的种类,所以白细胞胞核的边缘检测对临床标本的自动检测和分类具有重要意义.但是由于在成像过程中,常常混入随机噪声,这使得原始图像(把尺度空间中零尺度下的结果看作是原始图像)的边缘检测比较困难.小波分解是近年来较常用的方法,它以不同尺度下小波变换所获系数的局部极大值来确定图像的边缘.但是小波分解的方式固定,而正交小波包分解可以根据信号的特性灵活的选择分解方式,因而更加适合于图像的边缘检测.本文尝试利用正交小波包分解的方法实现白细胞胞核的提取并最终取得了比较好的效果.

The Algorithm and Implementation of Applying Wavelet Packets to Obtain the Leucocyte's Nuclear Edge

For the pathologist,the configuration of leucocyte's nu clears is one of the most important gist to observe and check.It is also the crux for the t echnology of leucocyte's auto classification.It is a key in distinguish ing all kinds of leucocytes images.So the edge detection of the leucocyte's nucleus is of great importance in the detection and classification of clinic samples.We use wavelet packet analysis to overcome the random noise in the image form process and imply frontier focusing to overcome the slight distortion after wavelet packet transform.And it works well.