Search

In this thesis, we apply Power Spectrum in fabric weave recognition, use power spectrum to analyze fabric image by image lightness features at definite illumination, and then extract fabric weave texture features. We can obtain fabric weave image power spectrum curve by Fourier Transform. From the curve, we can extract direction information of fabric weave, and get fabric weft density, then calculate weft Tex, weft number, weft repeats. Experiments prove the method practical and useful for fabric weave computer automatic analysis.

In this thesis, based on the weft organization features parameters analysis results, according to warp yarn arrangement features , apply correlation coefficient and autocorrelation function in weft weave texture feature recognition, establish weft cell image, analysis the weft cell image correlation coefficient, find out the number of weft circle, set up same kind weft cell image of the same phase of weft unit, calculate the warp brightness values of the same kind weft cell image, distinguish the weft points area and warp points area through luminance signal analysis, analyze the tissue points area changes between the adjacent similar weft cell, and determine the location and density of warp yarn. Experiments prove that the Fourier-transform method is feasible and has considerable accuracy.