Applications And Research Of Image Processing And Recognition In The Garment Industry

At present,digital sewing equipments are widely used in garment industry.The pattern making software is used and the patterns are made by hand in the process of pattern making.This means of pattern making has a low level of automation and intelligence,which leads low efficiency and precision of pattern drawing.This paper designs a software that can automaticlly extract and recognize the pattern in photos,which realizes automation of pattern making.The main research of this paper is that through digital image processing and recognition technology the outer edges of objects in photos are extracted and recognized as the basic graphics automatically from the photos shot in certain light conditions.First of all,the shot bitmap was imported and grayed.The algorithms that damage little edge are used on gray image filtering.The algorithms include median filtering and gray morphology filtering.After median filtering,the image's gray value range is recoveryed by gray stretch.Then,the outer edge of the target object in the result image is extracted after filtering.In this paper,improved Canny algorithm is used to extract the edges from the photos.The improved Canny algorithm uses Otsu method to compute the high gradient threshold,which can adaptively distinguish between edge of noise and edge of target object.After the edge extracting result's connecting,regions marking,thinning,branches removing,the noise's edge is removed and the target object's edge remains.Finally,the basic graphics on the target object's edge in the photos is recognized,the edges is recognized as line segments and curves.In this paper,line segment recognition uses least square linear regression method.Goodness of fit is introduced as fitting judgement of line segment recognition.The conner point which contains the main image shape information can be used as start point or end point of the edge's linear regression,so the conner point should be found first.Curve fitting uses parabolic curve spline interpolation method,the conner points are used as interpolation nodes.The algorithms used in this paper were test and verified by comparisons, experiment results show that the software reaches the goals of design.