中文摘要：

本研究针对复杂散焦的尿沉渣图像的精细分割,提出了首先使用小波变换和形态学处理消除散焦影响并进行图像的粗分割,然后根据粗分割得到的子图像的情况采用边缘检测或者自适应阈值分割的混合分割方法进行细分割,最后再采用剥离算法处理待分割的粘连重叠成分的分割。该方法不受散焦影响,充分利用了图像的多种信息,因此分割结果准确。实验结果表明,该方法对尿沉渣图像的分割有效且令人满意。

英文摘要：

To solve the segmentation of the complicated defocusing urinary sediment image, this paper proposed a new segmentation method. The method were conducted by following steps：（1） using wavelet transforms and morphology to erase the effect of defocusing and realize the first segmentation, （2） based on the result of first segmentation, using edge detection and adaptive threshold segmentation respectively, （3） using ‘peel off＇ algorithm to deal with the overlapped cells＇ segmentations. The method was not affected by the defocusing, made good use of many kinds of characteristics of the images, so it can get very precise segmentation. The experimental results showed that the method was effective for the multi-object defocusing image segmentation.