With the rapid development of information technology, the application media of visual identity design has been greatly broadened, and the requirements of dynamic variability and interactivity of vision have become higher and higher. The introduction of data as an element has brought new semantic endowments to visual design. To solve the problems of noisy image reconstruction and many outliers in traditional art design, this paper uses an improved phase correlation algorithm to reconstruct multimedia video images. In addition, based on the visual characteristics of human eyes, a multi-feature fusion viewpoint image quality evaluation algorithm is proposed. The simulation results show that the method adopted in this paper improves the uneven image texture in art design, and its real effect is greatly enhanced.