This paper concerned with the issue of image segmentation for intensity inhomogeneity, and a novel double regional evolution (DRE) based on level set method is proposed for image segmentation. The proposed DRE method introduces the local regional control term and the rectangle initialization contour, which can able to quickly segments uneven grayscale images and accelerates the curve evolution rate. In order to reduce the noises effects for image, the image segmentation method adopts Gaussian filter operator and convolution calculation. In particularly, DRE method develops a new potential penalty function as regularization term, which is embedded into the level set evolution equations to increases the constrained conditions based on the gradient flow conditions. Due to the potential function promotes the image contour evolutions with bilaterally extended, during the evolution, it is improved the image contours segment efficiency and numerical accuracy. Compared with the recent proposed methods based on level set, there are some superiorities of DRC method for image segmentation.