Study On The Identification Technology For Asphalt Pavement Crack Based On Digital Image Processing

As two kinds of the most important distress forms of asphalt pavement, fatigue cracking and rutting have become the main factor which influence the performance of the pavement. According to the regulation about the evaluation method of highway maintenance quality (try out) developed by ministry of transportation, in order to formulate the corresponding maintenance strategy, the department of highway maintenance should implement regular distress investigation about road surface condition. And the crack of asphalt pavement is visual form of pavement fatigue cracking and rutting, so it's a trend that the crack of asphalt road surface will become an important indicator. Traditional artificial fracture investigation method has not adapted to the demands of the development of highway, whose main disadvantages are slowly detecting, subjective influence, high risk of operation, and high cost, time consuming and disturbing traffic. Therefore, the development and research of automated pavement distress detect system is the current hot issue and difficult problem to be solved. The article will focus on some problems of pavement distress that exits in the crack detection algorithm, and study on the image enhancement, segmentation, crack measurement and classification, the integrated image of the research on pavement management system by using the principle and methods of digital image form. And according to the pavement distress image features, the key research is about pavement crack segmentation algorithm. The main research contents and results are as follows:In terms of image preprocessing, according to the characteristics of the pavement crack image, in order to highlight the image of directional fracture linear characteristics, extensions of single template, and constructing eight direction template. Put forward and realize the mask image smoothing preprocessing algorithm, and it shows its superiority by comparing with other methods in the experiments.The main study is focused on pavement crack automatic recognition algorithm based on digital image processing. First of all, we improve the OSTU threshold segmentation algorithm and put forward the within and inter distance between the threshold criteria. Secondly, we improve gradient wavelet transform and modulus maximum value image segmentation algorithm, put forward new wavelet image segmentation algorithm, and to be successfully used to the recognition of crack image with the use of wavelet technology, analysis of different wavelet base, decomposed levels, reconstruction strategies of pavement crack image recognition. Finally, we not only put forward the method about multi-scale top-hat transform by using the principle of mathematical morphology, but also achieve expected results in the road image segmentation experiments according by the types and the size of structural elements and the number of times about expansion operation.The paper proposes the methods of identification of linear and reticular cracks, recognition method of longitudinal and transverse cracks, and the identification method of massive cracks and crack based on the geometric feature extraction of the width, length, area, density and hole about crack. Compared with complex artificial neural network method, these methods has as many advantages as follows, the cost of calculation is low, no need for a large number of samples training. So the automatic classification of cracks should be carried out without human intervention.The paper put forward the integrated management information system of digital image of road. In the paper, the overall system framework, division and function of each module is introduced, the highway pavement design is analyzed, the highway pavement performance prediction and evaluation method is discussed, the highway pavement maintenance and maintenance plan is studied.