Improved method of fabric defects inspection based on local entropy

This paper presents an improved method of fabric defects inspection based on local entropy. The adaptive threshold was used for judging the defects of different pictures. It is good for reducing the error due to the change of the background and the subjective factor. The Tyler expand is also used to simplify the formula of the local entropy to accelerate the speed of calculation. In addition, noise in pictures is filtered out and the shape of the defects remains the same after the filtering by opening function of mathematical morphology. Final experiment results shows that this method has the advantage of high identification and good anti-noise ability.