| A huge conservation management issues became increasingly prominent highway network with the constant improvement of our highway network and the highway construction scale of growth increasingly flat. Pavement cracks were the main pre-form Surface Defects. It could maximizly prevent the loss of road assets if the scientific conservation methods were timely developed to avoid disease expansion.The traditional manual tests could not meet modern road speed and precision requirements for detection, and the automatic detection technology is imminent.Therefore, this article pavement cracks of automatic identification technology based on digital image processing were studied. The main work as follows:Combining the characteristics of pavement cracks, the linear gray stretch algorithm and Butterworth low-pass filter were used for image enhancement and filtering process, which was in order to further fracture identification provides a good image basis. And split the traditional global thresholding and local thresholding algorithm was insufficient, an adaptive thresholding method, histogram analysis for each 64 × 64 squares, each unit adopts automatically obtain appropriate threshold for image segmentation, a good image segmentation was got.Multi-directional variance denoising was proposed for the image of cracks noise directivity and dispersion of the local spots. It was removed according to the gradient noise ratio in the range according to crack direction variance of further use8-direction variance Denosing local noise removal,which got a good picture crack extraction.Based on support vector machine to achieve the image of cracks classification statistics, which will recognize automatically classified as transverse joints crack,longitudinal seam, crack and split block. Finally, based on Visual Stadio development platform, digital image recognition was integrated and classification algorithms cracks, cracks digital image recognition system was developed. |
PDF Full Text Request