A software program enables a user to track a scene or a spot on Earth from space (such as from the ISS) using an innovative algorithm. This robust and highly accurate software allows a scene to be tracked that can be not only the shifted version of a previous scene, but can also be a distorted one.
A user can choose an M×M-pixel subimage from a previous or older image frame captured by a camera from space, preferably close to the center of the scene to be tracked, where M is preferentially a power of 2, such as 128. Such a subimage, r(x,y), is referred to as a reference cell.
The user can choose N×N-pixel cell, S(x,y), from the current or newer image frame. This cell is called the test cell, where N>M, with N again preferentially a power of 2, such as 256. Then the user calculates the cross-correlation (CC) of the reference cell r(x,y) and the central M×M-pixel subimage, s(x,y), of S(x,y) using FFT (fast Fourier transform). The location of the CC-peak is determined by fitting a quadratic-curve to three points near and including the CC-peak in the x-direction, and doing the same in the y-direction. This is done analytically since there are three data points for three unknown parameters in such a fit. The test cell is shifted by the amount determined in the previous step to match it with the reference cell, using the Fourier-transform of the larger test cell, S(x,y), to avoid wraparound errors.
This work was done by Erkin Sidick of Caltech for NASA’s Jet Propulsion Laboratory. This software is available for commercial licensing. Please contact Dan Broderick at
This Brief includes a Technical Support Package (TSP).
Tracking a Scene on Earth from Space Using the Adaptive Cross-Correlation Algorithm
(reference NPO49156) is currently available for download from the TSP library.
Don't have an account?
Overview
The document discusses the Adaptive Periodic-Correlation (APC) algorithm developed for tracking scenes on Earth from space, particularly using images captured by the Extended-Scene Shack-Hartmann Sensor (ES-SHS). The research, conducted at NASA's Jet Propulsion Laboratory (JPL), highlights the algorithm's ability to accurately estimate shifts between images, which is crucial for monitoring moving targets from space.
The introduction outlines the limitations of conventional image registration techniques, which typically achieve an accuracy of about 1/10 of a pixel. In contrast, the APC algorithm has demonstrated the capability to provide high accuracy, achieving shift estimations of 0.01 pixel or less for images with good contrast. This is particularly beneficial for applications involving large images, such as those captured from the International Space Station (ISS).
The document presents several numerical examples to illustrate the effectiveness of the APC algorithm. For instance, in the first example, a satellite image is analyzed where a test cell of 256x256 pixels is compared to a reference cell of 128x128 pixels. The algorithm successfully estimated shifts of approximately [5.031, 3.836] pixels, with minimal estimation errors. Subsequent examples showed even better accuracy, with shift estimation errors decreasing to [0.006, 0.002] and [0.0006, 0.0013] pixels for images with improved contrast.
The research emphasizes the algorithm's potential for tracking moving scenes or spots from space, showcasing its application in real-time monitoring and analysis. The findings suggest that the APC algorithm can significantly enhance the precision of tracking systems used in various aerospace and Earth observation applications.
The document also includes references to previous works and acknowledges the support from NASA, highlighting the collaborative nature of the research. Overall, the APC algorithm represents a significant advancement in image processing techniques, offering promising solutions for high-accuracy tracking of dynamic scenes from space. The research underscores the importance of continued innovation in aerospace technology and its broader implications for scientific and commercial applications.
