Pavement crack identification method based on IOtsu-Dd algorithm

• Published in: PloS one Vol. 20; no. 5; p. e0322662
• Main Authors: Yang, Yang, Wang, Lin, Xiong, Qinghua
• Format: Journal Article
• Language: English
• Published: United States Public Library of Science 14.05.2025; Public Library of Science (PLoS)
• Subjects: Accuracy; Algorithms; Automatic vehicle identification systems; Computer and Information Sciences; Cracking; Cracks; Deep learning; Drone aircraft; Efficiency; Engineering and Technology; Highway construction; Highway maintenance; Identification methods; Image processing; Image Processing, Computer-Assisted - methods; Methods; Optimization algorithms; Pattern Recognition, Automated - methods; Pavements; Physical Sciences; Repair & maintenance; Research and Analysis Methods; Roads & highways; Unmanned Aerial Devices; Unmanned aerial vehicles; China
• ISSN: 1932-6203; 1932-6203
• DOI: 10.1371/journal.pone.0322662

Rapid identification of highway cracks is greatly significant for highway maintenance. In recent years, the use of unmanned aerial vehicles to collect images of road cracks for automatic recognition has become a topic of concern for many researchers. Based on this, to raise the accuracy and efficiency of crack recognition, a road crack recognition method based on unmanned aerial vehicle images and improved Otsu method is developed. Firstly, certain processing techniques are applied to the images captured by the unmanned aerial vehicle, such as grayscale and equalization, to reduce computational complexity and facilitate subsequent identification of image cracks. Subsequently, to improve recognition accuracy, the image is segmented and the Otsu method is introduced and improved. Finally, a pavement crack recognition model is constructed using damage density, achieving the extraction and recognition of pavement crack features from images. The experiment findings show that the raised recognition model has an average accuracy of 98.2%, a recall rate of 0.75, and an F1 score of 0.85 in crack recognition of unmanned aerial vehicle captured images. This denotes that the raised recognition model has strong effectiveness and high recognition accuracy, and the method can effectively recognize road cracks based on unmanned aerial vehicle images.