Detection Algorithm Research On The Surface Defects Of The Rebar

The appearance of the rebar surface complex characteristics, the author of thispaper, based on the theory of machine vision technology research, the test study on theappearance quality of the rebar and concrete by Matlab7.0combined with vc++6.0programming tools, realized the complex background of the appearance quality of therebar detection algorithm research. Joint of rebar front and side image processing, theappearance quality of the rebar is studied, the effect of the processing method designedthe subpixel edge location method based on projection of center of gravity, providesreference for rebar online detection.This paper studies the main work is as follows:(1) For rebar in common noise type, smooth denoising processing methods werestudied, through a variety of algorithms process of the comparison of the experimentalresults, final selection processing speed, good processing effect is median filtersmoothing method to suppress the rebar images contain noise.(2) To the need of rebar size calculation, respectively for rebar positive image andprofile image for different image processing process. For rebar profile image featureanalysis, made the image processing. According to the processing procedure, andanalyzes the rebar profile image binarization segmentation algorithm, by contrast,treatment effect dajin method (OTSU method) is chosen as the rebar profile imagebinarization segmentation algorithm; Using edge operator for edge extraction, chooseCanny edge operator for edge detection; After get the rebar profile image edge, the useof subpixel edge location method based on projection of center of gravity of rebarprofile image on the analysis and calculation, the size of the rebar image of cross ribheight and rebar diameter size.(3) Analysising of the characteristic of rebar positive image,to make the imageprocessing. Rebar are analyzed based on the processing flow, the algorithm of edgedetection, through edge operator for edge extraction effect, choose Canny edge operator for edge detection operator; After get the rebar profile image edge, edge contourtracking algorithm for rebar side the size of the image analysis and calculation, the finalresults.(4) Taking the way of classification based on decision tree method to classify therebar surface shape defect characteristics of typical pattern recognition,Theexperimental results show that this method has high accuracy.