Algorithm Study Of Automated Detection Based On Image Analysis For Bridge Substructure Distress

Bridge distress causes adverse effects on bridge in many aspects, such asdurability, mechanical wear, driving comfortableness, environmental protection, trafficsafety, etc. Traditional method based on artificial vision for bridge distress detectionhas a lot of shortcomings such as time-consuming, dangerous, costly, inefficiency, lowaccuracy and so on. In recent years, bridge distress detection based on digital imageanalysis has been developed greatly in highway maintenance field. However theautomatic detection algorithm is still not satisfying. Artificial vision method is themain measure for the later image processing.This paper is devoted to research the automatic detection algorithm includingbridge surface image enhancement, denoising, image segmentation, crackclassification, crack abstract and crack positioning.The main work is described asfollows.For image segmentation,based on the mathematical morphology, multi-directionmorphological structuring elements algorithm is proposed aiming to the differentdistributing characteristics between the noise and the disease pixel of the bridge image.For cracks classification, Based on different features of geometric shape,projection, crack distribution density and hole number are used to distinguish differentcrack. A pattern classifier based on RBF neural network is used to recognize differentcracks according to the geometrical shape differences of different cracks. Comparedwith other image segmentation algorithms, experimental result indicates that the imageedge is extracted, and noise is eliminated. Moreover, classification accuracy isachieved.In the key issues of the bridge parameters characteristic, it has been sorted cracksobjectives by the contour tracking algorithm, designed a effective algorithm of the netcrack by the calculate area, perimeter and shape parameters. Experimental results showthat the algorithm can accurately measure the single goal of the bridge the cracks in theimage parameters, and can effectively identify the crack.