全自动尿液图像识别技术

提出了一种全自动的尿液显微图像处理方法.提出一种新的图像预处理方法,通过拉伸图像中各个像素的灰度值与局部灰度值之间的差来增强图像中目标的边界,通过对局部灰度均值的非线性变换来消除图像中光照不均匀的影响.设计恰当的细胞神经网络（CNN）模板分割图像,采用形态学操作和对边缘链码序列的操作,分离黏连细胞,得到分割的最终结果.在获取目标区域后,提取目标的各种形态学参数和纹理参数,采用多个多层感知（MLP）网络分层次地对目标进行分类,得到全自动的处理结果.通过对大量临床尿液样本图像的测试,该方法获得了良好的分割和自动识别结果,并已经集成到全自动尿液图像分析系统中应用于临床,取得了良好的效果.

Automatic urinary image processing

A novel approach for automatic urinary microscopic image processing was presented. First, an image preprocessing method enhances the edges of objects by stretching the  difference between every pixel and the local gray mean value, and eliminates the disequilibrium of illumination by nonlinear transform of the local gray mean value. Second, a  suitable template of cellular neural network (CNN) is designed to segment the images, and the morphology operation and the boundary direction code operation are carried out to  get final results. Third, the shape and texture features are extracted from the objects. Finally, a two-layer classifier is built to get the last classify results of the  objects, and each layer is combined by multiple multi-layer perceptron (MLP) networks. The experimental results with many clinical urinary images showed that this approach  provides good boundaries and accurate object identification results. The related algorithms have been successfully integrated into automatic urinalysis systems and gained good  results in clinical applications．