基于自适应阈值和灰度共生矩阵的路面裂缝检测方法
Pavement crack detection method based on adaptive threshold and grey scale covariance matrix
针对传统路面裂缝检测方法在复杂场景下噪声敏感、特征单一及精度不足等问题，提出了一种自适应阈值滤波和灰度共生矩阵的路面裂缝检测方法。该方法集成高斯、中值及改进双边滤波算法，通过自适应权重分配策略动态优化噪声，结合形态学骨架提取与灰度共生矩阵（GLCM）纹理分析进行综合表征裂缝形态、纹理及空间分布特征，并采用激光雷达和三维点云实现裂缝长度、宽度等参数自动统计，采用美国马萨诸塞州公路损伤目标检测数据集验证所提方法。实验结果表明，该方法在噪声干扰条件下图像信噪比提升8 dB以上，裂缝检测精确率达96.78%，误检率降至3.8%，单帧处理时间仅1.2秒，效率提升3倍；点云配准误差小于0.15 mm，曲面拟合度优于95%。与传统高斯、双边、维纳、Gabor及导向滤波方法对比，验证了本文方法在复杂环境下具备良好的鲁棒性与识别精度。该研究为多源干扰场景下的路面裂缝检测提供了有效技术路径，并可拓展应用于桥梁、隧道等结构病害识别。
公路运输工程雷达电子技术应用计算技术、计算机技术
路面裂缝自适应阈值灰度共生矩阵形态学骨架
汪义娟,马海峰.基于自适应阈值和灰度共生矩阵的路面裂缝检测方法[EB/OL].(2025-09-19)[2025-09-28].http://www.paper.edu.cn/releasepaper/content/202509-23.点此复制
Aiming at the problems of noise sensitivity, single feature and insufficient accuracy of traditional pavement crack detection methods in complex scenes, an adaptive threshold filtering and grey scale co-production matrix pavement crack detection method is proposed. The method integrates Gaussian, median and improved bilateral filtering algorithms, dynamically optimizes the noise through adaptive weight allocation strategy, combines morphological skeleton extraction and grey scale symbiotic matrix (GLCM) texture analysis to comprehensively characterize the crack morphology, texture and spatial distribution, and uses LiDAR and 3D point cloud to realize automatic statistics of crack length, width and other parameters, and adopts the target of damage detection dataset of the state highway of Massachusetts in the United States to validate the proposed method. The proposed method is validated using the Massachusetts Highway Damage Detection Dataset. The experimental results show that the method improves the signal-to-noise ratio of the image by more than 8 dB under the noise interference condition, the crack detection accuracy reaches 96.78%, the false detection rate is reduced to 3.8%, the processing time of a single frame is only 1.2 seconds, which improves the efficiency by three times; the point cloud alignment error is less than 0.15 mm, and the surface fitting degree is better than 95%. Comparing with the traditional Gaussian, bilateral, Wiener, Gabor and guided filtering methods, the method in this paper is verified to have good robustness and recognition accuracy in complex environments. This study provides an effective technology path for pavement crack detection under multi-source interference scenarios, and can be extended to identify structural diseases in bridges and tunnels.??
Pavement cracksAdaptive thresholdingGrey scale covariance matrixMorphological skeleton
展开英文信息
评论