A new CNN deep learning model for computer-intelligent color matching

Abstract To accurately color the clothing of textile enterprises, many scholars have proposed intelligent color-matching models. However, these intelligent color-matching models still have problems with incomplete color-matching functions and low color-matching accuracy. To solve the above problems, this study optimized the convolutional neural network (CNN) algorithm using the genetic algorithm (GA), extracted data through the GA, and then processed the information using the CNN algorithm. Besides, based on the optimized algorithm, an intelligent color-matching model was constructed. To verify the performance of the proposed color-matching model, the superiority of the optimization algorithm (PSO) was first tested. The results showed that the accuracy of the PSO reached 98%, and the loss value of the algorithm function was only 0.02, significantly better than other algorithms. Further analysis of the performance of the intelligent color-matching model based on PSOs showed that the accuracy of the model’s color matching reached 98%, and the matthews correlation coefficient index of the model was 1. The performance of this color-matching model was superior to other models. In practical applications, the model had an average color difference of only 0.51, 0.49, and 0.47 for the three primary colors of red, green, and blue, with small color differences and high color-matching accuracy. From the above results, it can be seen that the intelligent color-matching model proposed in the study can improve the accuracy of color matching and reduce color difference, thereby improving the efficiency of clothing color matching in textile enterprises and promoting the development of the textile industry and other color-matching industries.