Spectral and spatial domain approach for fabric defect detection and classification

Abstract

Fabric inspection has an important role of preventing delivery of inferior quality product. The fabric defect inspection process is traditionally carried out with human visual inspection that proves to be insufficient and costly. Hence in order to reduce the cost and wastage of time, Computer vision based fabric defect detection is required. Robust and efficient fabric defect detection algorithms are required for inspection. This paper presents spectral domain and spatial domain approach for fabric defect detection. Spectral domain approach includes Fast Fourier Transform (FFT), Discrete Cosine Transform (DCT), wavelet and Gabor filter. Spatial domain approach includes two methods morphological operations and statistical operations.