A woven textile automated on-line flaw detection method

The present invention relates to an automatic machine line flaw detection method a woven textile, a fabric using two camera for acquisition of the first cloth piece pickup camera upstream region texture extraction, whether the initial detection of qualified products, defective downstream if the camera then pickup fabric texture extraction secondary partition, and defect recognition; two cameras shoot the same piece of fabric in a different location, to avoid wrongly as reflective interference caused due, identifying a secondary subregion after the initial recognition, greatly improve the recognition accuracy, while ensuring detection speed, take up less resources, able to adapt to online testing, non-stop testing. When the present invention, texture feature extraction, feature extraction method using the Markov model tree complex wavelet texture extraction were combined, compared to traditional texture extraction method enables rapid, efficient, accurate extraction of texture features woven textiles.