Defect detection algorithm based on gradient and multithreshold optimization

Classical edge detection algorithm cannot completely remove blur edge and lost sharp edge when it is used to process the defects of the timber. In order to resolve this problem, we propose a defect detection algorithm based on multi-threshold and gradient optimization. Firstly, through k-means algorithm (k=4), the mean threshold of module is required. Secondly, the image is segmented by 4×4 module, dynamic thresholds for each module are dynamically obtained; the gradient, the maximum modular value, the maximum difference of pixel value and the mean of multiple thresholds of modules are subsequently determined. Finally, the acquired modules are output and combined into a complete image, after median filter, optimized extracted image is formed. Through the subjective and objective evaluations, it shows that our algorithm improved the effect and quality of the image processing.