An Improved Watershed Segmentation Algorithm of Medical Tumor Image

Abstract

Aiming at the problem that the tumor is difficult to be accurately segmented due to noise and edge discontinuity in the chest breast MRI image, a medical image segmentation algorithm based on gradient edge detection and control marked watershed is proposed. Firstly, the mathematical morphology reconstruction filter is used to remove the small texture and noise of the MRI image of the breast tumor, and then the local minimum associated with the object is extracted from the low-frequency components of the gradient image, and they are formed into a binary marker image; The reconstructive technique is forced to be the local minimum of the original gradient image, the adaptive correction of the gradient image is realized, and finally the watershed segmentation is performed. The experimental results show that the algorithm can automatically and quickly extract the breast tumor region in the series of MRI images and realize the automatic segmentation of the region of interest in the medical image.