Identification Technology Of Superficial Hidden Voids Based On Features Of Ground-based Radar Images

In recent years,frequent ground collapses have caused great losses to society and economy,and also posed a huge threat to people's life and safety.As an effective means for detecting hidden voids in roads at present,Three-dimensional geological radar has the advantages of fast detection speed and intuitive image.Intelligent identification of underground hidden dangers based on 3-D geological radar images can effectively improve the detection efficiency and accuracy,and is of great significance to ensure the safe operation of urban roads.Based on the imaging principle of geological radar and the characteristics of cavity image,a 3D radar hidden danger imaging technology based on image recognition is proposed in this paper.Through this method,the feature analysis of GPR image based on standard volume model test is carried out;At the same time,the method is applied to the field cavity image recognition and analysis,and a three-dimensional volume calculation model is established.This paper mainly includes the following three parts:(1)3D radar hidden danger imaging technology based on image recognitionBased on the principle of GPR detection and data recording,the hyperbolic imaging model of underground point target based on effective detection width is derived;According to the amplitude ratio of the reflected wave between the boundary point and the vertex in the hyperbolic model,the formula for calculating the gray threshold of the hidden area and other areas in the gray image is given.Then the image is segmented with the help of binary method to realize the recognition of hidden hole area;Then,the outlier detection method based on KNN algorithm is used to detect and eliminate the abnormal interference region in the image,and the main range of hidden hole is extracted;Finally,the concave envelope algorithm based on the rolling ball method is used to obtain the two-dimensional boundary line of the cavity hidden danger area,and the three-dimensional point cloud image representing the cavity hidden danger body is obtained through the point cloud interpolation between the coordinates of the adjacent depth boundary line.(2)Feature analysis of GPR image based on standard volume model testFirstly,the relationship between the area of hidden danger boundary line and buried depth is analyzed.There is a positive correlation between the area of hidden hole boundary line and the depth of hole,and the radar image area will change approximately linearly when only the depth of hole is changed;Secondly,the relationship between the area of hidden image boundary line and the surface curvature of the specimen is analyzed.The greater the surface curvature of the specimen is,the larger the hidden image area of radar is.On this basis,the area difference calculation model?S based on the horizontal area S₀and surface curvature X of the specimen in a specific depth is established.The verification shows that the calculation model is reasonable under the occurrence conditions of the test environment.(3)On-site cavity image recognition and establishment of 3D volume calculation modelFirstly,based on the above radar image hidden danger range recognition and 3D imaging technology,the boundary line recognition and 3D imaging of four cavities are realized,and the analysis results of the perimeter,area and volume of the boundary line of all horizontal slice images are obtained.Then,the image analysis results are compared with the field re measured values.The main research contents include:(1)The deviation relationship between the parameters of the cavity boundary line and its average and maximum values and the field re measured values is analyzed;(2)The variation law between the area and perimeter of the cavity boundary and the irregularity of the actual cavity boundary is studied;(3)The variation law between the volume deviation EV and the area dispersion coefficient CVs of the four cavities is studied,and a three-dimensional volume calculation model is obtained.Through this model,the relationship between the software image analysis results and the field re measured values is established,which is of great significance to the efficient development of the road hidden cavity detection.