Investigation Into Measurement Of Fabric Wrinkling Based On 3-D Laser Scanning

With the development of economy,people have higher request about the quality of clothing.Fabric made in clothing,in the wearing process inevitably is wrinkling,which affects its appearance.So it is important to accurately test and appraise the fabric wrinkle resistance.The existing fabric wrinkle test method is difference from wrinkle situation under actually wearing.It can not really evaluate fabric anti-wrinkle ability in the process of actual wearing.In cases of this kind,this paper presents a new fabric wrinkle resistance test methods on simulating actual wearing wrinkle,and build the device which visually simulates the knees and elbows wrinkling.The characteristic parameters are extracted for 3 D and 2 D using 3-D laser scanning as well as image processing,and two methods of advantages and disadvantages are comparatively analyzed.Good correlation feature extractions are chosen to identify wrinkle grade using neural network technology.The research content and research results are as follows:(1)A device of simulating the human body knees and elbows wrinkling is set up,and the device can simulate static and dynamic wrinkles,and control wrinkle time,angle,and wrinkle shape of stimulated device is similar to wrinkle shape in wearing process.(2)The lab sets up 4 variable factors which can affect the fabric wrinkle degree,wrinkle way,time,number,loose quantity.The result show:for poor anti-wrinkle fabric,loose quantity has a big influence on fabric wrinkle form,and wrinkle way,time,number have little impact on fabric wrinkle form.For good anti-wrinkle fabric,4 factors has no impact on fabric wrinkle form.(3)The feature parameters are extracted based on 3 D laser scanning.The pearson correlation coefficient between contrast as well as average deviation value and wrinkle grade are greater than-0.8,and the pearson correlation coefficient Between index and grade from unit normal vector of unit normal vector direction value and wrinkle grade are greater than pearson correlation coefficient from the height direction.The pearson correlation coefficient between mean value of absolute form unit normal vector Z direction and wrinkle grade is 0.679.(4)The feature parameters,mean value and standard deviation of energy,entropy,contrast,correlation,are extracted based on gray-Level co-occurrence matrix.The pearson correlation coefficient between mean value and standard deviation of entropy and wrinkle grade are greater than 0.8,when image pixel is 100×150.(5)When 3-D and 2-D feature together to judge fabric wrinkle grade are combined,they can improve precision of forecast model.Precision of regression equation between the standard deviation of height direction from unit normal vector,entropy standard deviation and wrinkle grade is 87.5%,which is 6.5%greater than regression equation between standard deviation of height direction from unit normal vector and wrinkle grade.(6)RFB neural network is used to train and forecast wrinkle grade,and the accuracy of wrinkle grade forecast is 83.3%.The wrinkle grade of printed fabric have deviation with subjective assessment,because print make a big difference to eyes,which bring about the difference between subjective and objective evaluation.So we need contact subjective evaluation abut printed fabric.