Research On Pavement Crack Recognition Based On Image Processing

Cracks are the main form of pavement surfaces damage.Real-time monitoring cracks is greatly valuable for pavement maintenance and performance evaluation.The existing algorithms for crack detection are mostly based on the ideal condition and lack the adaptability to the complex environment,so it is difficult to meet the actual need of engineering applications.In view of the above problems,this thesis makes an intensive study on pavement crack detection in the following four aspects: image preprocessing,crack detection,crack quantification and assessment,as follows.(1)In the image preprocessing,firstly,the algorithms of shadow removal and lane line removal are applied to effectively remove the large area noise in the image background.And then,combining the image enhancement algorithm based on Retinex with the denoising algorithm based on the weighted least square filter can effectively sharpen cracks and retain the edge of cracks.(2)In the crack detection,an improved algorithm based on Histogram Contrast visual saliency detection is proposed in this thesis.Taking Precision,Recall,F-Measure as evaluation indexes,the thesis compares the improved algorithm with other detection algorithms,and makes qualitative and quantitative analysis.The experimental results show that the algorithm can effectively reduce the saliency value of back ground noise and suppress the dark noise.(3)In the crack quantification and assessment,an algorithm to quantify and evaluate the cracks by combining the crack rate with the width of them is proposed,according to crack damage assessment criteria and the different repair methods of different cracks type.The algorithm simplifies the quantification process,which can not only accurately assess the crack damage level,but also reduce the calculation amount and improve the efficiency.Research on pavement crack recognition technology based on image processing in this thesis after a large number of experiments can well adapt to the pavement crack recognition under the complex background,which is conducive to the reasonable assessment of pavement performance,the prediction of possible disease types of pavement,and the formulation of the corresponding maintenance scheme.