FORMED ELEMENTS SEGMENTATION OF URINE SEDIMENT IMAGE BASED ON GAUSSIAN MIXTURE MODEL

In order to solve the problems of urinary sediment images with various kinds, low contrast and include blurred borders, a formed elements segmentation method of urine sediment image based on Gaussian mixture model (GMM) is proposed. The edge intensity and area density were fused by using the constructed spatial constraint relationship to extract the effective area of the image. The core region enhancement method was used to enhance the effective area of texture features and construct the complete observation data. The spatial correlation of adjacent pixels was introduced into the local region to constrain the Gaussian mixture model. The conditional iterative algorithm was used to optimize the maximum posterior probability of the label field to realize image segmentation. The experimental results show that the proposed method could successfully improve the accuracy and completeness of image segmentation.