A Model for Segmentation and Distress Statistic of Massive Pavement Images Based on Multi-scale Strategies

Conventional distress detection method which dealing with each image through single algorithm and under single scale with lower efficiency. A robust and high-efficiency model for segmentation and distress statistic of massive pavement images which based on multi-scale space is proposed in this paper. It based on the facts that: (1) the crack pixels in pavement images are darker than their surroundings and continuous; (2) images associated with the same road section with the consistence of pavement texture structure. The proposed model contains three stages mainly: Image segmentation is implemented based on neighboring difference histogram method, then the weighted multi-scale based distress statistical is executed to get the crack index of the pavement images, in the end it separates the cracked images from massive images through the distribution of the crack index and achieved the objective of improving detection efficiency. Experiments results demonstrated that the proposed method can pick up the cracked images from massive pavement images correctly and effectively, and the time consuming is less than one third of the classical flow while the missing detection rate not exceed five percent. * LIU Xianglong, PhD Candidate, Tel.: +86-27-68778222, Fax: +86-27-68778043, E-mail: [email protected]