Pattern Recognition of Fabric Defects by Fourier Analysis

The construction of pattern recognition system for detecting fabric defects requires to solve the following problems : 1) How can we obtain the suitable parameters of the defect itself? 2) What is the analytical metohd for evaluating these parameters? 3) To what extent can we explain the relation between the recognition sytem and ordinary visual sensory property? A streakiness defect was discussed to answer the above questions. Fourier analysis predicted that streakiness defect was influenced by the statistical deviations of the irregularities of a yarn coarseness (σd) and of a yarn arrangement (σp).Experimentally, the combination of the above two deviations, √σ2dR2/2+2σ2p= 0.11 and the ratio of a yarn coarseness to a yarn spacing R = 0.71, were obtained by using a commercial fabric in the recog nition system.