The Research On 3D Pavement Crack Detection Algorithm

With the completion and application of pavement, some early pavement diseases will appear. These early pavement diseases can be caused by series of technical issues in the design, construction and maintenance, and also can be caused by repeated traffic load of vehicles and continuous effects of climate and environment. These early pavement diseases will lead to the gradually weakened pavement function if not repair timely. Therefore, an effective pavement maintenance plan for early pavement diseases has important value and practical significance. The formulation of pavement maintenance plan depends on the feature data of the pavement diseases. Pavement crack diseases are the main form of the pavement early diseases, so it is important to research an effective pavement crack detection algorithm. In this paper, analyses of existing pavement crack detection algorithms were made, the specific research content of this article and the key technology need to be focus on were established based on these analyses.In this paper, according to the characteristics of 3D image acquisition system and 3D pavement data, the pavement crack identification algorithm based on 3D data were studied. The crack identification and feature parameters extraction were completed. according to the pavement disease feature extraction and detection requirements, combined with the digital image filtering processing technology, a 3D pavement crack identification method based on dual-phase scanning detection was put forward. The paper used the specific method to deal with the noise data points. This specific method combined the standard deviation average filtering with the cascade morphological filtering algorithm based on composite structural elements. Then cracks could be identified through the dual-phase scanning detection based on the average method of curve fitting. A crack image was obtained based on horizontal single-phase scanning detection for pavement 3D data matrix. The other crack image was obtained based on vertical single-phase scanning detection for pavement 3D data matrix. Next the intersection for two crack images and region growing operation were processed. Crack was identified by dual-phase scanning detection. Next crack feature parameters were extracted based on the identified crack image.For the 3D pavement crack detection algorithm in this paper, a lot of indoor and outdoor experiments were developed, results show that the method can identify cracks and extract crack feature parameters better. The algorithm given in the paper has good stability and high accuracy by accuracy rate curve assessment analysis. The paper provides an important guidance for pavement maintenance management decision-making and pavement 3D morphological reconstruction.