Image segmentation is a fundamental and challenging problem in image processing and often a vital step for high level analysis. Being CV based models have the unsatisfactory segmentation results and inefficient curve evolution against weak boundary and intensity heterogeneous images because of the inappropriate initial contour and unbalanced using the local and global information of the image. Based on the study of the edges and local/global contrast of the image, in this paper, we proposed a stable active contour model of image segmentation. Firstly, a new automatic initial contour choosing algorithm has been proposed which may improve the evolution efficient to a large extent compare to the human chosen initial contour. Besides, this algorithm may also improve the accuracy of the segmentation regions. Secondly, based on the study of the local binary fitting (LBF) model, local/global information fitting (LGIF) model and edge-flow based active contour model, we proposed a gray and contrast guided active contour model. In this model, we use gray and contrast information of the image as a decision standard to balance the local and global information. Finally, based on the above two algorithms, we construct a new image segmentation framework. The experiments show that our algorithm is less dependent on the parameters compare to the other models. On the other hand, this algorithm may also improve efficient of curve evolution to a large extent. Extensive experiments on synthetic and real images are provided to evaluate our method, showing the segmentation of the blurry boundary and intensity heterogeneous images may achieve more accuracy results.