Research on a New Fabric Defect Identification Method

A new method of recognition the fabric defects features is proposed for alleviating the difficulties of extracting to complicated fabric defects features. First, fast Fourier transform and self-adaptive power spectrum decomposition are performed. Sector-regional energy of spectrum is extracted and its mean and standard deviations is calculated as fabric features. Then, the spectral energy distribution was objected to direction Y, and local peak was extracted as defects recognition features after objection. Fabric defects recognition using the proposed method shows a high performance in the on-line detection.