Pavement Disease Image Automatic Classification Of Research And Analysis

The automatic identification and classification of image-based road surface distress has been a challenge work on image processing and pattern recognition, which plays a great role in guiding on highway maintenance projects. The background of my paper is a Jiangsu Province Fund Project of Nanjing University of Science & Technology: A research on road surface distress detection, classification and measurement. This paper explores some main characteristics of the road surface distress image and does research on the automatic identification and classification based on images of road surface distress.Based on the pavement distress binary image, the main research of the paper is going to find some feature extraction methods about road distress, and propose structure feature targeted to a variety of road distress through detail and sound analysis of road distress images, including the linear feature, the intensity feature and the regional statistical feature, which give a very good description of the various types of surface distress characteristic shape with the obvious effect of classification. Meanwhile the paper emphasized the fractal feature of road surface distress by calculating the fractal dimension of the road distress binary images. Generally, the box-counting dimension and information dimension can be used to distinguish three broad categories of pavement distress: unidirectional cracks, network cracks and pits.Finally, a classifier is also designed by using BP neural network, the input is a eigenvector which is the integration of structure feature and statistic feature of a road surface image and the output includes six categories of distress, the classifier gets a better recognition to five main types of road surface distress.