Research on Digital Image Recognition of Surface Cracks in Plate Steel Based on Sobel Operator
Based on the characteristic that the color depth of surface cracks on steel is higher than that of adjacent areas,this paper studies the automatic identification of steel surface cracks by image processing.Using Sobel operator to detect image edges and identify surface cracks in steel plates.Using a binocular stereo vision system as the hardware input for image acquisition,analyze the specific methods of image distortion processing and camera calibration for image acquisition.In terms of specific algorithms,this article converts the obtained crack images from RGB format to YUV data,and uses Sobel operator to perform convolution calculation to obtain image gradients.The gradient values are further subjected to root mean square operation to obtain pixel grayscale values;At the same time,the crack grayscale threshold is trained multiple times based on the type of steel and surface characteristics to obtain the steel crack discrimination threshold under different conditions.Compare the threshold with the calculated image grayscale value,assign a value to the color channel based on the comparison result,and then obtain the pixel points of the crack.Further interpolate the black pixels to obtain a continuous steel crack curve.
surface cracks on steel platesimage recognitionSobel operatorgraph data analysis
NETL
NSTL
万方数据
