A multi-scale convolutional and color-adaptive approach for sensory enhancement in cultural and creative product packaging

This study addresses the critical need for enhanced visual appeal in cultural product packaging by proposing a novel multi-scale convolutional neural network (MCCNN) with adaptive color enhancement. Unlike existing methods that struggle with uneven lighting and detail loss, our approach innovatively combines laser-based 3D feature fusion with illumination-aware enhancement to overcome these limitations. The method extracts multi-level visual features from packaging images through scale transformation and feature fusion, constructing a laser-based 3D multi-scale feature fusion model to achieve image preprocessing and noise reduction. Furthermore, by employing block matching and fuzziness detection techniques, a visual constraint model is established to effectively extract features from blurred regions and detect image block information. In terms of image enhancement, the integration of illumination compensation and adaptive dehazing techniques addresses issues such as image fogging and detail loss during brightness adjustment, thereby improving image quality and color richness. Experimental results demonstrate that the proposed method achieves a 90.62% completeness rate in 3D reconstruction of product packaging images, with an average design time of less than 5.3 s. Additionally, the color enhancement module shows outstanding performance, with a color enhancement effect of 94.99%, an image fitness value of 1.0148, and an information entropy of 78.96%, effectively enhancing image contrast and visual quality. This research offers new insights and technical support for the intelligent sensory design of cultural and creative product packaging.