Natural image matting is an important algorithm on image processing to extract the foreground objects from the background image. This paper proposes a Markov random field(MRF) model based approach to natural image matting with complex scenes. The image is manually

divided into three regions:fore region

back region and unknown region

which is segmented into several sub regions. In each sub region

we partition the colors of neighboring background or foreground pixels into several clusters in RGB color space and assign matting label to each unknown pixel. Each label is modeled as an MRF and the matting problem is then formulated as a maximum a posteriori(MAP) estimation problem. Simulated annealing is used to find the optimal MAP estimation. The better results can be obtained under the same user interactions when the image is complex. Results of natural image matting experiments performed on complex images using this approach are shown and compared in this paper.