Method for automatic on-line detection of detects of woven textile

The invention relates to a method for automatic on-line detection of defects of a woven textile. According to the method for the automatic on-line detection of the defects of the woven textile, fabric collection is conducted by means of double cameras, grain extraction is carried out on the whole piece of fabric which is in the shooting area of the camera in the upstream firstly, whether the fabric is a qualified product is preliminarily detected, zonal secondary grain extraction is carried out on the fabric which is shot by the camera in the downstream if the fabric is unqualified, and detect recognition is carried out at the same time. According to the method for the automatic on-line detection of the defects of the woven textile, due to the fact that the same piece of fabric is shot by the two cameras at different positions, misjudgment caused by interference such as light reflection is avoided; due to the fact that the zonal secondary recognition is carried out after the preliminary recognition, the recognition accuracy is greatly improved, the detection speed can be guaranteed at the same time, a small number of resources are occupied, and on-line detection and non-stop detection can be achieved. Due to the fact that grain extraction is carried out by means of the feature extraction method with dual-tree complex wavelets combined with a Markovian model, compared with a traditional grain extraction method, the grain extraction method adopted by the method has the advantage that the textural features of the woven textile can be extracted rapidly, effectively and accurately.