Human body auto-extraction using graph cuts

Grabcut is an Interactive segmentation algorithm, which is employed to extract the foreground of an image, but it works wrong frequently when the pixels around the foreground are similar to the pixels of the background. We advance a new cost function to correct the mistake after we analyze the reason why it works wrong. Experimental results show that the new cost function outperforms the original one. Moreover, based on the new cost function we propose, we advance an algorithm of human body auto-extraction, which can be used to extract human body from background without interaction of user.