The Evaluation Of Pilling Grade Of Fabric Based On Images Analysis Technology

Along with more and more requirements what people ask for the quality of fabric, people pay more attention to the pilling problem of fabric. During the process of fabric quality controlling and exporting inspecting, both need to evaluate the anti-pilling performance of fabric. The usual method that people apply for is comparing the standard photos. This method belongs to subjective evaluation method which contains many man-made factors, so can't evaluate the degree of fabric pilling accurately. The thesis introduces digital image processing technology, then does a serial analyses and disposals with five grades' standard pilling image of combing wool fabric, extracts the eigenvalue, and sets up the relationship between character parameters and the grades of pilling.The brief of the thesis is introduced as follows:1. A serial pretreatments of the standard image of combing wool fabric is introduced, including cutting of the image and removing the fabric's texture. Mainly introduces the method of removing the fabric's texture by using the transform of Fourier in the area of frequency, removes the bright dots and strips what are corresponding with the textile texture information by applying medium-value filter in the area of frequency, and discusses what size of medium-value filter is the best.2. The methods of image segmentation are introduced and discussed. Two kinds of methods are introduced and compared, one is based on boundary, and the other one is based on threshold. The method based on boundary applies Roberts arithmetic operators, Sobel arithmetic operators, Canny arithmetic operators and Log arithmetic operators, the method based on threshold applies self-adaptation threshold and Gaussian threshold, compares the two methods, and mainly studies the feasibility and rationality of Gaussian segmentation method, and then filtrates yawp points which are less than three pixel of two-value images.3. The pilling feather's indexes are extracted. The relationship between the number of pilling, the total area of pilling, the maximal area of pilling, the standard difference of piling area, the contrast degree, the optical volume, the density distributing of pilling and the pilling degree is being analyzed and discussed in detail. Finally chooses the number of pilling, the total area of pilling and the optical volume as character parameters. And then does unitary transaction and sets up the vectors of character parameters. 4. The methods which can evaluate the grade of pilling are compared and the method of minimum distance evaluation is being chosen. Finally chooses three sets of images of actual pilling fabrics, transacts the uneven light of images. And then evaluates the actual fabrics pilling degrees by using the same whole set of methods.