Intelligent Simulation Design and Application of Garment Digitization Based on Virtual Image Recognition Technology

Abstract

Realizing the digital design of clothing is an effective means to improve the design efficiency and effect, for this reason, this paper designs a virtual simulation design assistance system based on virtual image recognition technology. Wavelet Fourier descriptor and LDA technology are utilized to extract features and decrease the dimensionality of the clothing style image, while the extreme learning machine is utilized as a classifier to categorize the clothing style image. Based on the classification results, a virtual simulation model of the garment is constructed and stored in the parts library for the convenience of designers to call and design at any time to complete the construction of the design assistance system. The application of this paper’s image recognition method results in an overall style recognition rate of 94.8%, an average recognition time of 3.2 milliseconds for each sample, and an average Fourier wavelet descriptor classification accuracy of 97.5%. The size error is controlled within ±2.5%, and the mean customer satisfaction scores for all the finished sweaters are higher than 2. This paper proposes a style map recognition method that achieves high accuracy, reduces time consumption, produces finished products with minimal size errors, and increases customer satisfaction with the designed styles.