论文部分内容阅读
为准确识别道路土基病害以避免路面塌陷事故的发生,采用探地雷达对城市道路进行检测。针对城市非硬化道路和硬化道路土基病害出现的一般规律,首先通过维纳滤波器对探地雷达图像进行一维滤波,在滤波过程中,对增益函数进行倒谱域内的平滑处理,进而得到增强的雷达图像;然后,采用傅里叶变换、小波变换等信号处理方法,得到探地雷达图像在不同变换域的结果,并通过构建傅里叶核函数、多项式核函数、克罗内克核函数等不同的核函数字典,对探地雷达图像在不同变换域的结果进行匹配追踪,比较在不同核函数下的核匹配追踪序列;最后,通过比较不同核匹配追踪序列的差异,分别找出对道路土基水害和空洞病害类型敏感的核匹配追踪序列,通过该序列识别城市道路土基病害的类型。利用基于相关系数的病害度量算法比较核匹配追踪序列,判断城市道路土基病害发生的区域,并通过算法识别非硬化道路与硬化道路的地下病害。结果表明:病害类型及范围与实际情况相符,水害和空洞均能得到有效识别;通过5组城市道路探地雷达探测数据集验证了算法的有效性,识别准确率达到了99%以上;利用核匹配追踪算法处理探地雷达图像有助于城市道路土基病害识别,可减少路面塌陷事故的发生。
In order to accurately identify the road-based diseases in order to avoid the occurrence of pavement collapse accidents, GPR is adopted to detect the urban roads. Aiming at the common law of earth-based diseases in urban non-hardened roads and hardened roads, the Ground Penetrating Radar image is first filtered through Wiener filter. In the filtering process, the gain function is smoothed in cepstral domain, Then, the signal processing methods of Fourier transform and wavelet transform are used to obtain the results of GPR images in different transform domains. By constructing the Fourier kernel function, the polynomial kernel function, the Kronecker kernel Function and other different kernel function dictionary to match and track the results of GPR images in different transform domains and compare the kernel matching trace sequences under different kernel functions.Finally, by comparing the differences of different kernel matching trace sequences, The sequence of nucleus matching tracking that is sensitive to the types of water-borne and void-borne diseases on the road can be used to identify the types of earth-based diseases in urban roads. The disease matching algorithm based on correlation coefficient was used to compare the kernel matching tracking sequence to determine the area where the soil-based diseases occurred in the urban road and to identify the underground diseases of the non-hardened roads and hardened roads by algorithm. The results show that the type and the range of the disease are in accordance with the actual situation, and the water damage and the cavity can be effectively identified. The validity of the algorithm is verified by five groups of urban road GPR detection datasets, the recognition accuracy is above 99% Matching tracking algorithm to deal with ground penetrating radar images is conducive to the identification of urban road-based disease, can reduce the occurrence of road surface collapse.