论文部分内容阅读
分割算法被广泛用于公路路面裂缝的检测中,然而低抗噪性是这类方法的主要问题。提出一种采用双树复小波变换的路面裂缝检测方法,将双树复小波变换和直方图方向梯度相结合。该方法首先用双树复小波变换对路面裂缝图像进行子带分解,然后对各子带图像进行直方图方向梯度矩阵计算,阈值化后确定裂缝边缘。通过实验证明,与传统边缘检测方法相比,本文所提出的方法目标识别度高,抗干扰能力强,且准确率较高。
Segmentation algorithm is widely used in the detection of road pavement cracks, however, low noise immunity is the main problem of this kind of method. A pavement crack detection method using double-tree complex wavelet transform is proposed, which combines the double-tree complex wavelet transform with the histogram direction gradient. In this method, the fractal images of the pavement are first subband-decomposed using double-tree complex wavelet transform, and then the histogram direction gradient matrix is calculated for each sub-band image. After thresholding, the crack edge is determined. Experiments show that compared with the traditional edge detection method, the proposed method proposed in this paper has high target recognition rate, strong anti-interference ability and high accuracy.