Automatic Detection System Of Road Surface Disease Improvement And Related Algorithms

With the development of expressway, the technology of expressway surface disease inspection is also urgent to be enhanced. The background of my paper is the project of Nanjing University of Science & Technology, which named "N-1 Style Intelligent Road Surface Disease Inspection Vehicle".Our goal is to improve the performance of the intelligent disease detection subsystem and make it more reliable and useful.After researching the technology of image analysis and the existing algorithms on disease detection, I provide a disease detection process. This process applies a fast preprocessing algorithm. It can pick out the suspicious images. These images may contain diseases and will be analyzed carefully later. In the preprocessing step, images are selected by the histogram character of its subimages. In the detail analysis step, the wavelet transform is adopted to extract the high frequencies. After processing these high frequencies, do an inverse transform, and get the outline of cracks. Then, use the threshold segment and connective region expansion methods to get the whole cracks and calculate the statistics of the diseases.Because a lot of the images have no diseases and won't be picked out in preprocessing step, the images needed to be analyzed in detail decrease a lot. Fast algorithms are adopted, and the detection system avoids a lot of multiplication and division acts. The new system is more effective and the veracity of the result is enhanced.To meet the needs in practice, the thesis provides a design of a simple and useful distributed system. It can engage several computers to run the disease detection system. Another diseases recognition method based on the artificial neural networks is introduced briefly at the end of the paper.