Detecting the surface defects on hot rolled steel sheets using texture analysis

At present the detection of surface defects on hot rolled steel sheets is most trivial problem facing by the steel Industry. There are few methods available for detection of surface defects and the most popular method is texture analysis. In this paper, we highlighted the extraction of the texture features using a three level 2-D Haar wavelet transform, and training the Artificial Neural Network (ANN) classifier to detect the presence of surface defects on hot rolled steel images. The algorithm was tested with 45 defects free and 55 defective images and the results prove that this method gives 100% defect detection. This approach is very promising in checking the presence of surface defects with low resolution and non-uniform lighting images. This work has been implemented using wavelet and neural network toolboxes in MATLAB.