Research And Achievement Of Recognition Of Road Distress In Automatic Check Of Road Imformation

Research target: 1. Establish a new fuzzy enhancement algorithm, by using Pal-King fuzzy enhancement algorithm as study basis, for the damaged road image which is shot by multi-functional automatic highway traffic data collection vehicles in the speed of 70km / h. This algorithm can quickly enhance the images and make the enhancement images nearly reflect the true distribution of distress of the road and reduce the burden of manual identification work.2. Establish a new fast image segmentation algorithm, by using the study of the traditional image segmentation techniques. The image which is segmented by this algorithm can exactly reflect the distress of the road.3. Achieve the automatic recognition of damage types of road, by using a new feature vector extraction method and a cluster analysis approach to complete the effect of self-learning.The main research contents: 1.Analyze a variety of road damaged image enhancement technology for multi-purpose high-grade highway truck which can automatically collect traffic data and find the lack of these algorithms. In the actual application, though the edge and region and texture of image are fuzzy, the description of the results of image processing is fuzzy. So this paper focuses on the classic Pal-King fuzzy enhancement algorithm, find the insufficient of the algorithm to improve the program in order to establish a new fuzzy enhancement algorithm.2.Study on the classic image segmentation algorithm, analysis of the value of the reasons for poor results, find a algorithm which can apply to the image collected by multi-purpose high-grade highway truck which can automatically collect traffic data. By analyzing the binary image, establish a new feature vector extraction method, achieve automatic recognition.The results of research: Based on the classic Pal-King algorithm, establish a fuzzy enhancement algorithm combined with image sub-block which can fast cope with road damaged image and remove the interference of shadow and closely reflect the true distribution of crack binary image.2.Based on the study of traditional road damage image segmentation algorithm, according to the same region between the pixel gray-scale changes in gentle and undulation and statistical variance is small, The undulation and statistical variance of the edge pixels in the region between the gray value is large, present a new image segmentation algorithm based on the value of standard deviation and mean value of gray-scale images.3. According to the new algorithm, using a new feature vector extraction method, achieve high accuracy automatic classification purposes.