An Effective Technique for Detecting Defecting Parts of Fabric in Digital Image

Inspection and detection of fault parts of the fabric are one of the major issues in the textile industry. Currently, detection of the faulty parts of fabric by the human vision system, which detect very less percentage and decreases the quality of the fabric. So, there is a need to develop a new real-time technique that consists of a computer vision system that works automatically to detect all fault parts of fabric and increase the quality of the fabric. In this paper, we proposed a method to detect the defects in the fabric. To achieve this objective, Fast Fourier Transform and Cross-correlation techniques, i.e. linear processes are first realized to scrutinize the building uniformity land-escapes of the fabric picture in the incidence area. To progress the competence of the system and speechless the tricky of uncovering errors, an extra thresholding process is realized using an equal collection filter. Through this sieve, the system is able to detect only the definite or real defects and high point their meticulous proportions. Also, Recurrent Neural Network and Artificial Neural Network are used to classify the faulty and non-faulty parts of the image. Results shows that proposed method is performed better as compare to baseline methods.