Application of high-resolution satellite remote sensing technology in safety infrastructure emergency monitoring of traffic infrastructure

With the continuous improvement of spatial resolution, remote sensing information has brought convenience to the renewal of geo-information databases. High-resolution remote sensing images have been widely used in extraction of road network information.

Transportation infrastructure extraction based on high-resolution optical images

With the continuous development of related technologies, the technology integration of high-resolution remote sensing technology and traffic emergency security has gradually become one of the hot topics in recent years. In order to better promote the application of high scores technology in the future traffic field and the transformation of achievements, this paper proposes a road extraction using high resolution optical remote sensing and Synthetic Aperture Radar Interferometry (InSAR) technology based on road infrastructure. And monitoring technology.

With the rapid development of science and technology, the extraction technology and extraction algorithm for traffic network information has developed rapidly. With the continuous improvement of spatial resolution, remote sensing information has brought convenience to the renewal of geo-information databases. High-resolution remote sensing images have been widely used in extraction of road network information.

1. Establishment of a typical road texture database

The road network extraction technology based on the convolutional neural network algorithm usually needs to use training data to establish a network model. The establishment of typical road characteristics database is firstly to extract, classify, and statistically analyze typical remote sensing samples using their features such as morphology, structure, and connectivity, and then covers all typical road feature units as much as possible based on statistical results. The number will directly affect the algorithm's recognition and extraction accuracy.

2. Road network automatic identification algorithm

In recent years, with the appearance and development of deep learning technology, autonomous learning has become possible. The increase in the number of network layers also makes neural network classification ability stronger. The image regions are classified by the convolutional neural network, and the classification results of the pixels can be obtained. Then the classification results are regarded as binary images. The final extraction results can be obtained by screening the identification results by analyzing the size of the binary image intercommunication area. .

Subgrade Detection and Warning Based on Radar Image

The principle of using SAR image to obtain surface feature information lies in the difference of backscatter of different objects. When the internal structure of the local table changes, the information received by the satellite sensor will also change. Based on this principle, multi-temporal SAR images can be widely used in the deformation investigation of transport infrastructure.

1. Surface deformation analysis based on InSAR technology

Synthetic Aperture Radar Interferometry (InSAR) is an interferometric measurement mode of active microwave imaging sensors. It performs interference processing on an image with a certain angular difference and correlation in the same area, detects the phase difference, and converts it according to a certain geometric relationship. Finally, it will achieve the acquisition of terrain height data in the observation area. In order to ensure the accuracy of data, radar images based on it usually have a spatial resolution of more than 3 meters, and more than 7 scenes of multi-phase images. In engineering surveys, in order to guarantee the accuracy of data, the permanent scattering (PSInSAR) technique can also be used. PSInSAR technology has high requirements on data quality. If there are few natural permanent scattering points in the study area, a corner reflector can be set up as a supplement.

2. Image resampling and matching processing

Different from the gray information of optical remote sensing images, InSAR analysis results can provide intuitive surface deformation information, but the visualization effect of SAR remote sensing data is far less than that of optical images. That is, although it can obtain the deformation of a spatial coordinate point, it cannot be intuitive. To see what this point is. Therefore, it is usually necessary to match the road image extracted from the optical remote sensing image and the InSAR result to achieve the superposition and complementation of the two information. However, due to the differences in observation time and imaging modes between optical remote sensing image and radar image, when data fusion is performed, two kinds of images need to be spatially matched first, and the accuracy of matching will also directly affect the final data analysis and data mining. The reasonableness and accuracy. To solve this problem, an effective fusion method is: 1) establish a standard grid within the target area; 2) perform a bilinear difference between the optical remote sensing image and the radar remote sensing image based on the established standard grid, for different spaces Registration of the resolution image; 3) Remove the low confidence observation point in the InSAR result, and perform the matching and fusion between the InSAR result to the optical image data according to the registration mapping relationship in the previous step.

3. Subgrade condition monitoring based on road area

After the information fusion between the images is completed, a mask image is generated using the normalized road image, and the ground deformation information of the target road area is extracted in combination with the normalized deformation data, and subsequent data analysis and excavation are performed to realize the road. The extraction and monitoring of the status of the facilities, such as the loss of road usage, post-disaster traffic capacity, and analysis of road network connectivity in the region.

Compared with traditional artificial means, remote sensing data can not only greatly reduce human and material costs, but also can carry out large-scale data analysis and mining, which greatly promotes the development of related research. For the transportation industry, the effective use of remote sensing technology to service traffic safety emergencies is an important direction for the development of remote sensing applications.