Research Of Pavement Texture Detection And Recognition Method Based On Image Processing

With the rapid development of Chinese highway, the asphalt pavement is playingmore and more significant role in highway transportation system. Through a largenumber of surveys for traffic accidents in asphalt pavement, researchers found that theprobability of traffic accidents will reduce greatly when other conditions remain basicallyunchanged and only improve pavement skid resistance. Therefore, it is necessary to studythe characteristics of surface texture on asphalt pavement in order to reveal therelationships between the skid resistance and pavement texture.The traditional approaches to obtain the pavement texture information are includingsand patch method and laser method. However their disadvantages are obvious for thereason that the accuracy of the measured data collected by sand patch method is stronglyinfluenced by artificial factors. Additional, the laser method has poor economicalefficiency because of its expensive equipment. Thus, the photometric stereo methodbased on four light sources is employed to collect the pavement texture information.Firstly, the camera is applied to acquire the photos of pavement in different lightconditions. Secondly, the pavement normal vector is calculated by Matlab software. Atthe last we take advantage of tangent theory to recover the three dimensional data ofasphalt pavement. The photometric stereo method is relative cheap compared to the lasermethod. Moreover, the three dimensional morphology recovered by this method is closedto the real asphalt pavement. So it is feasible to use the photometric method to collect thetexture information on asphalt pavement.A hybrid optimization, which combined genetic algorithm and sequential quadraticprogramming, is applied in this paper for the purpose of solving the low accuracy issuecaused by traditional method in the progress of calculating fractal dimension of asphaltpavement. Additional, a laser rangefinder made by our own is used to testify theadvantages of the hybrid optimization.The result showed that photometric stereo method is relative cheap compared to the laser method. Moreover, the three dimensional morphology recovered by this method isclosed to the real asphalt pavement. So it is feasible to use the photometric method tocollect the texture information on asphalt pavement. In addition, on the basis of threedimensional model and laser data, the factors after optimization is better than the factorscalculated by traditional method in terms of fitness, approximation degree ofinterpolation curve and the accuracy of fractal dimension.