The Design And Realization Of The Pavement Distress Detection System

With the rapid development of pavement traffic industry, people are putting forward higher requirements on the quality and maintenance of the pavement surface. The monitoring method based on the visual inspection is difficult to have scientific and objective evaluation for pavement surface distress. So it's urgent to develop and design the automatic pavement distress detection system.This article is mainly research about the image processing of the pavement distress detection system. The development of the pavement distress detection system at home and abroad, and the processing on the pavement distress detection image are introduced. The processing includes image preprocessing, feature extraction and distress image classification.Image preprocessing is the key of the pavement distress detection system. This article gives a transformation method from true color bitmap to gray scale image. A reliable guarantee for feature extraction and distress image classification is provided by analyzing and comparing sorts of algorithm, selecting the appropriate image processing algorithms. In the transformation process, in order to outstand the characteristic of pavement distress image, the operationgs to remove the noise, the grey value average and the grey value outspread as well are carried on.Feature extraction is the basis of image analysis. BP neural network classifier which uses sub-block variance, multiple threshold value segmentation, image overall average gray level as parameters, is adopted. This non-linear classifier combines the advantages of sub-block variance and multiple threshold value segmentation to make up for theirs deficiencies.Distress image classification is the main purpose of the system design, which its accuracy affects directly identification rate of the automatic pavement distress detection system. According to the calculation results of connected domain number, the judgment of the distress image type can be completed.