Visual edge feature enhancement of product appearance design images based on improved retinex algorithm

Under the influence of complex factors such as lighting, color distortion, and suspended solids, there is a problem of losing edge feature information and blurring edges in product appearance design images. In order to improve the clarity and visual effect of product appearance design, a visual edge feature enhancement method for product appearance design images based on an improved Retinex algorithm is proposed. By using a color correction method based on depth of field estimation, the blue tone of the product appearance design image is removed, and color correction and contrast are applied to the product appearance design image. Improve the Gray Wold algorithm and design an edge attenuation compensation method to solve the problem of edge color attenuation under noise interference, and obtain clearer product appearance design images. On the basis of clarity processing, convert the original RGB image into HSV. On the basis of the Retinex model, multi-level decomposition of brightness is carried out, and different filtering parameters are set to obtain multiple illumination and reflection images with different scale information; Using exponential function and Sigmoid function to process reflection images and illumination images separately, reducing external interference on images of different scales, and solving the difficulty of enhancing images with uneven illumination, high noise, low illumination, and loss of details. At the same time, adaptive nonlinear correction is applied to the saturation component, and the corrected saturation, brightness, and hue are fused and converted into RGB, expanding the edge grayscale feature information in various spatial domains. Improve the weights of traditional bilateral filtering methods, reduce the depth difference between information at different scales, and enhance the visual edge features of product appearance design images. The experimental results show that the proposed method enhances the image with a PCQI of 1.033, an IQE of 0.610, an IQM of 1.830, and an information entropy higher than 0.7. The above data proves that this method has a high richness of edge feature information after image enhancement, significantly improving the visual edge feature enhancement effect of product appearance design images.