Research on the Intelligent Color Matching Algorithm Integrating Visual Aesthetics

In recent years, AI-assisted intelligent design has gradually emerged, with color matching methods based on deep learning technology becoming a focus of current research due to their ability to enhance color matching efficiency. However, color matching is constrained not only by physical laws but also by physiological and psychological factors. Therefore, establishing a color matching model that aligns more closely with visual perception based on deep learning technology has become a key challenge in current research. This paper addresses the issue of a lack of visually perceptive data-driven algorithms in image translation models for color matching by proposing an intelligent color matching algorithm that integrates visual aesthetics. The algorithm first utilizes eye-tracking technology to obtain a stream of visual aesthetic data, which then drives the Pix2Pix image translation model, followed by quantitative analysis using the aforementioned color matching evaluation methods. Experimental results indicate that, compared to image translation models that do not incorporate visual aesthetics, the quantitative analysis metrics and the peak signal-to-noise ratio of images improved by 0.096 and 6.12 dB, respectively. Additionally, in terms of subjective evaluation, the overall preference level increased by 9.28%, bringing it closer to the effects of ground truth palettes. This paper systematically analyzes the physiological and psychological attributes of color matching schemes, utilizing visual aesthetic score to drive the Pix2Pix image translation model for intelligent color matching, and validates its effectiveness and advancement through constructed multidimensional color matching evaluation metrics. This has significant guiding implications and application value for color matching in product design and artistic creation.