Research On Subjective And Objective Evaluation Of Visual Effect Of Distorted Visual Illusion Clothing Image

Distorted illusion is a phenomenon in which structural elements such as regular points,lines,and surfaces are perceived as distorted structures.It is widely used in clothing design to increase the interest and artistry of clothing.Designing different black and white stripes or geometric patterns that cannot focus on simple clothing styles,combined with the human body curve and the clothing dividing line to produce the illusion of distortion,can highlight the artistic effect of the clothing.The visual illusion patterns derived from the same prototype will show different visual effects on the clothing when the direction and size change,giving people different visual feelings.Therefore,visual illusion patterns are rationally used in clothing design,which can not only modify the shape,but also complement the clothing structure.At present,researches on visual illusion clothing mostly focus on the application of artistic design and modeling,and pay less attention to the subjective and objective visual effect of visual illusion clothing images on people.The lack of evaluation indicators and evaluation models for visual illusion clothing images hinders the application and development of visual illusion clothing.This article starts from the patterns of distorted optical illusion,and the distorted optical illusion patterns of different rules are screened and designed.The CLO 3D virtual fitting software is used to simulate the dressing effect and obtain images of visual illusion clothing.Through Kansei evaluation and objective eye tracker experiments,the influence of various distorting costumes on the subjects is comprehensively analyzed,to explore the relationship between subjective and objective evaluation parameters of visual perception of clothing image.The image features of the experimental samples are obtained by image processing technology,and the prediction model is established by neural network to provide data support for the design and application of the distorted visual illusion clothing images.The specific research content and results are as follows:（1）Acquisition of visual illusion clothing images:First,106 visual illusion patterns were determined,and one of them was selected to change the line ratio,rotation and twist angle and color parameters for secondary design,and 192 patterns were obtained,totaling 298.In the CLO3D virtual fitting software,the visual illusion pattern is filled as the texture of the fabric to simulate the wearing effect of the visual illusion clothing and obtain the experimental sample skirt images from different angles.（2）Objective eye tracker experiment:Distorted optical illusion patterns with regular changes are selected as experimental samples,and are divided into three groups of eye movement experiments with different distortion angles,different line widths,and different colors.Five indicators including pupil diameter,first fixation time,first saccade range,total fixation time,and fixation times are extracted,and combined with the heat map for group analysis,it is concluded that the design parameters of the pattern have a significant influence on the objective eye movement data.（3）Subjective Kansei evaluation:Irregular visual illusion pattern clothing images were subjectively evaluated and scored.The pattern of visual illusion that changes regularly is perceptually analyzed to obtain three Kansei evaluation factors of angle factor KN,line factor KX,and Color factor KY_a.And combined with objective eye movement data to summarize.It is concluded that the design parameters have a significant impact on the subjective evaluation and Kansei factors of the subjects.（4）Image feature extraction and model construction:The image processing technology is used in Mat Lab software to calculate the coarseness F_crs,contrast F_con,directionality F_dir,line-likeness F_lin,regularity F_reg and roughness F_rgh of the sample image based on the Tamura texture feature algorithm as the visual illusion clothing image Objective characteristics.The influence of pattern design parameters on texture features is analyzed,as well as the relationship between texture features and eye movement data.The Linear regression model and MLP neural network model of subjective and objective parameters of visual perception are constructed.The MLP neural network model can more accurately predict the subjective perception of distorted visual illusion images.