Research On Dressing Fit Based On Wrinkle Image Recognition

With the rapid development of e-commerce technology,online garment consumption has become an increasingly popular consumption model.In the process of online shopping,there is a contradiction between"Virtual Behavior"?online shopping?and"Physical Objects"?garment?.The most direct reflection of this contradiction is the fit of garment.In short,the fitness is an important factor affecting the appearance and comfort of garment.It is also one of the most important factors affecting the sales of the garment.The purpose of this paper is to analyze consumers'visual perception and wrinkle recognition based on the image in online shopping environment,to improve the evaluation of the dressing fit in the virtual environment,and reduce the error between the purchase pre-judgment and the physical object when the purchase is made.In particular,the pre-judgment error in the fit of the dressing,thereby reducing the return rate of the garment product.Therefore,this paper does the following research.Firstly,the influence factors of fit visual perception stimulation in dressing image are analyzed,the design experiment is tested,the data analysis is carried out by statistical method,and a visual perception evaluation model of dressing fit VTF?Visual to Fitting?is established.The experimental values of visual stimulus were tested by wrinkle,fabric performance,garment style,looseness of main parts,brightness and shade of color,and color reflection in the dressing image.Pearson correlation was used to analyze the degree of correlation between them and fit of dressing.Then factor analysis is used to extract four main factors:wrinkle visual factor,fabric color factor,garment looseness factor,garment style factor.Finally,based on regression analysis,a multivariate linear regression model,VTF model,is constructed for visual fitting perception V and four principal factors,and U-test is used to verify the accuracy of the model.Thus,a subjective evaluation model of visual perception of dressing fit based on image is established.Secondly,an objective evaluation model ITW?Image to Wrinkle?for the recognition of wrinkle elements is established by using computer Image processing technology.Based on the image gray index,the GCF threshold segmentation algorithm is designed,the changing trend of the gray curve is analyzed and fitted,the optimal threshold is extracted,and the target segmentation for the image is segmented.Canny edge detection algorithm is used to identify the wrinkle shape.According to the wrinkle shape and gray distribution,three quantitative indexes of wrinkle are proposed:wrinkle number,wrinkle regularity and wrinkle unevenness.In the evaluation of the wrinkle regularity,three morphological modes are defined innovatively to reflect the wrinkle trend:Point-Gathering?PG?,Line-Gathering?LG?,and Parallel-Line-Gathering?PLG?.Finally,the relationship between the wrinkle indexes and the dressing fit is analyzed.The objective evaluation of dressing fit based on wrinkle image recognition is realized by Matlab R2014a programming.Finally,an example simulation and verification of the established visual perception model VTF and the wrinkle image recognition model ITW are carried out.The development conception of corresponding intelligent evaluation software is proposed.The L₉?3,4?orthogonal experiment table was used to design orthogonal experiment with three factors:garment looseness,garment style and fabric color.Three levels of combination of each factor were designed for orthogonal experiment design.Nine sample garments were selected accordingly.9 pieces of sample garment are tested in the physical garment experiment to test the sense of fit score of the sample garment after the actual dressing,the visual perception evaluation value of the image taken after the sample dressing and the objective evaluation value of the wrinkle recognition of the image of the sample dressing.This paper analyzes and verifies the unity and complementarity of the two models of dressing fit evaluation based on wrinkle image recognition,then propose the development idea of the fit intelligent evaluation software.