Research And Implementation On Fast Image Segmentation Method Based On Graph Cuts

Image segmentation is the key issue in pattern recognition, computer vision and image processing and occupies an important position in image engineering. In recent years, image segmentation based on graph cuts is a popular method. Its core idea is to construct an energy function and minimize the energy function.By researching and analyzing Grabcut algorithm based on graph cuts theory, we find that the algorithm efficiency is greatly reduced because of using the whole pixels to iterate and estimate parameters. Through analyzing the algorithm, we can know that there’s no need to use all the pixels. We can achieve the goal by selecting a few representative samples. So, we proposed a fast image segmentation method based on wavelet transform and graph cuts theory. First we can process the image by the wavelet transform, and then use the low-frequency image’s pixels which can represent the main features as the sample points of GMM parameters estimation. Followed by the use of Grabcut algorithm to determine the established GMM parameters, so we can cut the original image and the get the target image.A large number of experiments show that the proposed algorithm can guarantee the quality of image segmentation, but also greatly improve the efficiency of image segmentation.