Abstract
For real-time monitoring of natural disasters, such as fire, volcano, flood, landslide, and coastal inundation, highly-accurate georeferenced remotely sensed imagery is needed. Georeferenced imagery can be fused with geographic spatial data sets to provide geographic coordinates and positing for regions of interest. This paper proposes an on-board georeferencing method for remotely sensed imagery, which contains five modules: input data, coordinate transformation, bilinear interpolation, and output data. The experimental results demonstrate multiple benefits of the proposed method: (1) the computation speed using the proposed algorithm is 8 times faster than that using PC computer; (2) the resources of the field programmable gate array (FPGA) can meet the requirements of design. In the coordinate transformation scheme, 250,656 LUTs, 499,268 registers, and 388 DSP48s are used. Furthermore, 27,218 LUTs, 45,823 registers, 456 RAM/FIFO, and 267 DSP48s are used in the bilinear interpolation module; (3) the values of root mean square errors (RMSEs) are less than one pixel, and the other statistics, such as maximum error, minimum error, and mean error are less than one pixel; (4) the gray values of the georeferenced image when implemented using FPGA have the same accuracy as those implemented using MATLAB and Visual studio (C++), and have a very close accuracy implemented using ENVI software; and (5) the on-chip power consumption is 0.659 W. Therefore, it can be concluded that the proposed georeferencing method implemented using FPGA with second-order polynomial model and bilinear interpolation algorithm can achieve real-time geographic referencing for remotely sensed imagery.
Highlights
With the advancement of technological, remote sensing (RS) images are becoming more widely used in natural disasters monitoring and positioning [1,2,3]
This paper proposes a georeferencing method based onfield programmable gate arrays (FPGAs) with the optimized second-order polynomial equation and bilinear interpolation scheme
Georeferencing is a key step in geometric correction which aims to establish the relationship between image coordinates and ground coordinates through various function, such as collinearity equation models (CEM), polynomial function models (PFM), rational function models (RFM), and direct linear transformation models (DLT) [10,11,12,13,14,15]
Summary
With the advancement of technological, remote sensing (RS) images are becoming more widely used in natural disasters monitoring and positioning [1,2,3]. This paper proposes a georeferencing method based onfield programmable gate arrays (FPGAs) with the optimized second-order polynomial equation and bilinear interpolation scheme. Georeferencing is a key step in geometric correction which aims to establish the relationship between image coordinates and ground coordinates through various function, such as collinearity equation models (CEM), polynomial function models (PFM), rational function models (RFM), and direct linear transformation models (DLT) [10,11,12,13,14,15] These traditional algorithms were developed for serial instruction systems based on personal computers (PC). The algorithms of RS data processing generally map quite nicely to multi-processor systems composed of clusters or networks of CPUs, these systems are usually expensive and difficult to adapt to the on-board RS data processing scenarios For these reasons, the specialized integrated hardware devices of low weight and low power consumption are essential to reduce mission payload and obtain analysis results in real-time.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
