To solve the problem of color gradient filling of complicated shape in its stretching trend, a Shape Main Skeleton Color Gradient Filling Algorithm (SMSCGFA) was proposed by using shape main skeleton. Based on visual salience estimating vector, the main skeleton was extracted from a shape by using whole selecting and local geometric optimization. Some important methods in SMSCGFA were studied that included extracting skeleton using Constrained Delaunay Triangulation (CDT), and extracting skeleton path by double stacks. Gradient filling color in main skeleton was computed, and the whole shape gradient filling color was completed by local main skeleton color filling information. The experiment results show that the optimized skeleton path ratio reduces to 5.5%, and more skeletons with redundant branches are eliminated, and SMSCGFA satisfies subjective visual perception in shape stretching trend compared with color linear filling.