Research Of Infrared Image Sequences Mosaic In Pavement Damage Evaluation

As the bridge of materials transport between regions, road tracffic plays animportant role in the rapid development of modern society. As time goes by, the roadshould be upheld for its worsing situation. In particular, the devris flow and landslidemay threaten the safety of the road traffic. In order to keep the traffic unhindered andthe materials transported smoothly, the pavement damage evaluation must be carried outrapidly and help the repair work accomplished immediately. Therefore, it is veryimportant to gain the damage information of pavement for rush repair.For the research of fast evaluation of pavement damage in night or low lightcondition, this paper take images of pavement by Infrared Thermal Imager, applys themethod of inforared image mosaic to the fast evaluation so that evaluating system hasanti-interference and hidden capabilities.Firstly, the method of infrared image mosaic is summarized. The key point ofinfrared image mosacic for pavement damage evalution is introduced. The attribute ofinfrared pavement image is studied intensively considering the way of imaging. Thenon-uniformity correction and inverse perspective transform are executed before imageregistration, so that the quality of the images is improved, the model of transform issimplified and the evaluation of the transform matrix is more stable.Secondly, the registration algorithm of Harris based on corner detection and SURFbased on blotch detection are analyzed theoretically, the way of image transformationand estimation of transform matrix is fixed. According to the characteristics of theexperimental images, a feature extraction method based on detection window is used tocut down the number of invalid features and reduce the calculation time. Parameters ofHarris are optimized, and a bi-direction matching strategy is proposed to improve theproperty of registration. Beside, through setting dynamic threshold of SURF, not onlythe correct matched number is enough, but also the property and the stability orapplicability of the SURF is improved. The experimental results show that the workabove is effective.Lastly, in order to enhance characteristic strength and improve the robustness ofinfrared image mosaic for pavement, a method to enhance infrared image feature isbrought out. It’s feasibility is analyzed theoretically, and experimental results show themethod improves images quality, robustnessof feature points matching and significantlyincreases matched number. By comparison of coordinates in pavement map betweenimages with feature enhancing and images without feature enhancing, it is indicated thatthe method effectively improve precision of infrared image mosaic for pavement.