论文部分内容阅读
利用频域中傅里叶变换投影定理,提出一种新的离散脊波实现算法,应用于高速公路局部线性裂纹的检测取得较好效果。详细阐述了离散脊波的实现步骤以及对标准图像进行脊波变换的模拟结果,并提出拉东(Radon)变换重建原图像的基本条件。上述定理应用于复杂背景下的路面检测,结合直方图均衡化算法消除背景噪声;选用基于样本估计的阈值方法对脊波分解的各层系数进行处理去除随机噪声。选用不同的重构系数进行计算,得到脊波变换后重构图像的信噪比优于二维小波变换(低频大于20dB)以及二维小波变换加魏纳滤波变换(平均大于3dB)。通过图像的二值化处理提取局部线性裂纹,其分辨力极限达到2mm精度。
Using the Fourier transform projection theorem in frequency domain, a new discrete ridgelet algorithm is proposed and applied to the detection of local linear cracks in expressway. The implementation steps of discrete ridgelet and the simulation of ridgelet transform of standard image are described in detail. The basic conditions of Radon transform reconstruction of original image are proposed. The above theorem is applied to the pavement detection under complex background, and the histogram equalization algorithm is used to eliminate the background noise. The thresholding method based on sample estimation is adopted to process the coefficients of each layer of the ridgelet decomposition to remove the random noise. With different reconstruction coefficients, the signal-to-noise ratio of the reconstructed image obtained after ridgelet transform is better than that of two-dimensional wavelet transform (low frequency greater than 20dB) and two-dimensional wavelet transform plus Weina filter transform (average greater than 3dB). The local linear crack is extracted through the binarization of the image, and the resolution limit is up to 2mm.