This paper analyzes the computer-aided art design system based on the image aesthetics measurement, and proposes a new image salient area detection algorithm. After improving the original method, the new algorithm obtains a new adaptive pre-segmentation method, which can adaptively set the number of blocks according to the complexity of the image content, making the segmentation more uniform and reasonable. A comprehensive image aesthetics design model is calculated, which includes two parts: aesthetic classification and score prediction. The image features extracted by this model include low-level visual features, high-level aesthetic features and regional features. Using the quadratic programming optimization function to solve the optimal grid composition position point.