Introduction - If you have any usage issues, please Google them yourself
This procedure is based on the support vector machine image segmentation problem. Through the support vector machine to the picture of the sampling point for training, and then the whole picture to predict. Images can be divided into two parts: foreground and background.