摘要
为了从复杂的路面环境中快速检测出路面裂缝并进行分类,提出一种多尺度矩阵滤波的路面裂缝检测方法。该方法基于TI公司TMS642DSP平台,可以实现大量复杂数据的快速实时处理,通过多尺度滤波处理路面图像,有效进行图像预处理,消除噪声影响,突出路面裂缝特征。通过Hessian矩阵的特征值和特征方向提取裂缝特征,实现裂缝生长方向的跟踪,确定裂缝像素始末位置,从而确定裂缝大小,同时提出裂缝合并算法对小的不连续裂缝进行合并,最后根据裂缝曲率对裂缝进行快速分类。实验结果表明,该方法可以实现路面裂缝准确、快速的检测及分类,抗噪声能力强,图像分割精度高,漏检率和错检率很低,可以满足工程应用要求。
In order to quickly detect pavement cracks and classify them from complex pavement envi- ronment, a method based on multi-scale matrix filtering is proposed. The method which based on TMS642 DSP platform from TI can process a large number of complex data with high speed in real- time, and effectively process pavement images from multi scales to eliminate the noise, then highlights the characteristics of pavement cracks. Characteristics of pavement cracks can be extracted by analyzing the feature and feature direction of Hessian matrix, then growth direction can be traced and the size of it can be determined. At the same time, the small discontinuous cracks are combined through the crack combination algorithm. Finally, the cracks will be classified according to the crack curvature. Experimental results show that the method proposed in this paper can quickly detect andclassify the pavement cracks with high accuracy, it also has strong anti-noise capacity, high image segmentation accuracy and low miss rate and error rate, so it can meet the engineering application requirements.
出处
《液晶与显示》
CAS
CSCD
北大核心
2016年第8期778-783,共6页
Chinese Journal of Liquid Crystals and Displays
基金
国家自然科学基金(No.61472267)
江苏省普通高校研究生科研创新计划项目(No.KYLX15_1311)~~
关键词
裂缝检测
滤波
曲率
图像分割
crack detection
filtering
curvature
image segmentation
