Used For The Pavement Damaged Image Mosaic Technique

Highway plays an important role in promoting social and economic development. With the rapid development of national economy and growing of traffic volume,pavement appears different degree of damage. It will directly influence the vehicle's safety and speed. Pavement maintenance has become the highway department's important work. Because of the limited range of the camera, it can only get parts of the pavement damaged images on one time. However,with the image mosaicing technique's rising, it can mosaic the partial overlapping pavement damaged image into a complete crack picture. So it can formulate rational maintenance program. Based on synthetic analysis of the research results in the field of image mosaicing and the research status, study deeply in pavement damaged image mosaicing. The specific content as follows.Firstly,study in the feature point extraction algorithm and give a method of the gray threshold adaptive selection. Due to the pavement image acquisition car get the pavement damaged image affected by illumination uneven.Use the fixed gray threshold to extract the feature points.It is easily leak or wrong extraction. This method is that according to the region gray of the edge points adaptively calculate the appropriate gray threshold when it iterates through every edge points. Experiment proves that this method increases the accuracy of the feature point extraction.It can better apply to extract the feature point of pavement image.Secondly, present a method of pavement image feature point rapid extraction. The characteristics of the pavement damaged image is that a smaller proportion of the pavement damaged image is crack. Aiming at the characteristics of the pavement image, The method of this paper presenting is that detects the edge of two images firstly.So it can narrow the scope of the feature point extraction. Then put the edge points into the improved adaptive threshold feature point extraction algorithm to effectively extract feature points.Thirdly, Improved the matching point away wrong algorithm. This algorithm is not directly random sampling in all the feature points.First of all,the matching point set is ordered by the correlation coefficients before sampling, thus it reduces the number of iteration and quickly searches for the optimal point set. Finally, through the experiments verify the correctness and effectiveness of the algorithms, and experimental results are analyzed and compared.The experimental results prove that the methods in this paper have good practical value in actual pavement image mosaicing.