Back to Search
Start Over
Graph Matching Based Image Registration for Multi-View Through-the-Wall Imaging Radar
- Source :
- IEEE Sensors Journal. 22:1486-1494
- Publication Year :
- 2022
- Publisher :
- Institute of Electrical and Electronics Engineers (IEEE), 2022.
-
Abstract
- Among various types of sensors, through-the-wall imaging radar (TWIR) is widely used in concealed targets detection and urban environment perception. Especially, multi-view TWIR has been received more and more attention in recent years for its ability to provide more accurate position estimation and reduce the blind areas. However, when exploiting the multi-view TWIR to detect targets in the complicated indoor environment with an unavailable global positioning system, the relative position of the radars is hard to be obtained and the acquired non-uniform images can not be aligned automatically. To deal with these problems, an image registration approach based on graph matching for multi-view TWIR is proposed in this paper. In this approach, we exploit an important feature of multi-view TWIR images: the invariant relative position of targets in different views. Based on this characteristic, the acquired images are modeled as complete graphs, and graph matching is employed to acquire the relative position parameters of the radars and register the images. Compared with the classic method, the proposed method can overcome the critical challenges of invalid local features as well as high-intensity ghosts, and achieve better performance for multi-view TWIR image registration. The effectiveness of the proposed method is evaluated via both simulations and real data tests.
- Subjects :
- Exploit
Computer science
business.industry
ComputingMethodologies_IMAGEPROCESSINGANDCOMPUTERVISION
Image registration
Position (vector)
Feature (computer vision)
Radar imaging
Global Positioning System
Computer vision
Artificial intelligence
Electrical and Electronic Engineering
Invariant (mathematics)
business
Instrumentation
Urban environment
Subjects
Details
- ISSN :
- 23799153 and 1530437X
- Volume :
- 22
- Database :
- OpenAIRE
- Journal :
- IEEE Sensors Journal
- Accession number :
- edsair.doi...........754cbe63c7dfd121254099d292316360
- Full Text :
- https://doi.org/10.1109/jsen.2021.3131326