Research On Algorithm Of Segmentation And Recognition Of Urinary Sediment Images

Urinary examination is to use microscope examination or other analytical instrument to analyze and check on visible component of urine, such as red blood cells, white blood cells, epithelial cells, casts and crystallization. It plays an important role in diagnosis and differentiation of kidney diseases, urinary tract diseases, circulatory system diseases and infectious diseases. The traditional microscopical inspection which is watched by one eye not only increases doctors' burden but also brings up artificial error. Furthermore, the image can't be transformed and processed. These problems can be overcome by urine sediment analyzer and computer processing. In this paper, identifying the components in images, such as urinary sediment image enhancement, segmentation, feature extraction and recognition are researched.In this paper, several major research contents are as follows:Because of the big ingredients and small ingredients have different characteristics, and poor discrimination between object and background and complicated defocusing in image. Firstly, the wavelet transforms are used to erase the effect of defocusing. Then, morphology method is used to locate urinary sediments in the images and the urinary sediment sub-images are obtained. Finally, based on the characteristics of the sub-images, wavelet transforms and two-dimensional entropy threshold are employed respectively to segment urine sediment visible components. Experimental results show that the proposed method can segment urinary sediment images effectively and precisely.In order to integrally and accurately describe urinary sediments, 24 feature parameters based on shape, statistics and texture are selected to describe urinary sediments, which lay a foundation and prerequisites for urinary sediment images automatic recognition.SVM (Support Vector Machine) is introduced into recognition and classification of particles in urine sediment. One-vs-rest strategy is used to solve the problem of multi-class pattern recognition. Two-level-classifier is designed to recognized urinary sediments. The kernel function uses the radial basis function and grid-search method is used to select the parameters of SVM. The experimental results show that SVM method has higher recognition rate of urine sediment classification.