基于支持向量机的尿液细胞图像识别分类研究

Urine cells image recognition research based on support vector machine

为提高尿液细胞进行识别分类的效果,分析和比较了在RGB和HIS两种不同色彩坐标系统下使用支持向量机对尿液细胞进行识别分类的效果,分析和比较了使用色彩特征参数与空间特征参数进行综合识别分类尿液细胞的效果,提出使用网格搜索交叉验证法对支持向量机的参数进行优化。实验结果表明,提出的HSI颜色参数、空间参数、网格搜索交叉验证优化选择参数相结合的方法在尿液细胞识别分类中效果良好。

To increase the identifying and classifying results of the urine cells. The identifying and classifying results of support vector machine （SVM） on the urine cells under RGB and HSI color coordinate system is analyzed and compared. The result of classifying combined with color and space characteristic parameters is listed. The grid search with cross validation is used to optimized the parameters selection in support vector machine （SVM）. The experimental results show that the combined methods with HSI color parameter, the space parameters and the grid search with cross validation optimization is better one for urine cells recognition and classification.