Density Detection Method of Three-Dimensional Tubular Woven Fabric Based on Fourier Transform

In the production of three - dimensional tubular woven fabric, online and real - time density detection is crucial. A low density can compromise the straightness rate, while a high density impacts the efficiency of the circular weaving machine and escalates business costs. In order to solve the problem of large image noise and high real-time requirement in the detection of the density of a hollow 3-dimensional tubular fabric, the frequency domain-based image density feature analysis algorithm is studied and a Fourier transform-based frequency domain feature enhancement algorithm is proposed. Initially, the Fourier transform is employed to convert the fabric image into the frequency domain. Subsequently, the autocorrelation algorithm is utilized to intensify the weft characteristics of the tubular woven fabric. Finally, the inverse Fourier transform constructed the weft spacing d in the spatial domain. Through experimental verification, it is determined that the optimal cut-off frequency scaling factor for low-pass filtering to eliminate noise is 0.02. When the number of detection iterations reaches four, the system can meet the accuracy requirements, with a detection period of approximately 420 ms. Through experimental comparisons with the boundary curve extreme point method and similar Fourier transform algorithms, the algorithm proposed in this paper demonstrates higher accuracy. Moreover, when the images are contaminated by noise, the method in this paper exhibits excellent robustness and stable output performance. This method has been verified and applied on actual production lines, where the detection efficiency and accuracy are greatly improved, enabling the automation of the detection process.