Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Detection of Moving Objects in Earth Observation Satellite Images

Eric Keto, Wesley Andres Watters

TL;DR

The paper tackles detecting moving objects in push-broom Earth-observation imagery where each scene is a mosaic of sequential exposures lacking precise timestamps. It analyzes two motion signatures—the color ghosts in composite RGB images and positive–negative pairs in differenced bands—using a Planet Labs SuperDove PL22 scene to infer motion. A timing-restoration framework is developed, deriving the color-band offset $\Delta t_{color}=\frac{N_y \mu}{2\pi R_\oplus \omega}$ and velocity estimates $v_i=\frac{\Delta p}{\Delta t_{color}+a}$ with $a\in\{0,\pm\Delta t_{camera}\}$, enabling velocity measurements despite mosaicking. The results show detectable motion for cars, boats, and aircraft, with velocity accuracy limited mainly by pixelization and altitude–speed ambiguity, and the work points toward automated motion-detection software for large-area Earth observation, with broader relevance to planetary-science and search-for-extraterrestrial-origin research.

Abstract

Moving objects have characteristic signatures in multi-spectral images made by Earth observation satellites that use push broom scanning. While the general concept is applicable to all satellites of this type, each satellite design has its own unique imaging system and requires unique methods to analyze the characteristic signatures. We assess the feasibility of detecting moving objects and measuring their velocities in one particular archive of satellite images made by Planet Labs Corporation with their constellation of SuperDove satellites. Planet Labs data presents a particular challenge in that the images in the archive are mosaics of individual exposures and therefore do not have unique time stamps. We explain how the timing information can be restored indirectly. Our results indicate that the movement of common transportation vehicles, airplanes, cars, and boats, can be detected and measured.

Detection of Moving Objects in Earth Observation Satellite Images

TL;DR

The paper tackles detecting moving objects in push-broom Earth-observation imagery where each scene is a mosaic of sequential exposures lacking precise timestamps. It analyzes two motion signatures—the color ghosts in composite RGB images and positive–negative pairs in differenced bands—using a Planet Labs SuperDove PL22 scene to infer motion. A timing-restoration framework is developed, deriving the color-band offset and velocity estimates with , enabling velocity measurements despite mosaicking. The results show detectable motion for cars, boats, and aircraft, with velocity accuracy limited mainly by pixelization and altitude–speed ambiguity, and the work points toward automated motion-detection software for large-area Earth observation, with broader relevance to planetary-science and search-for-extraterrestrial-origin research.

Abstract

Moving objects have characteristic signatures in multi-spectral images made by Earth observation satellites that use push broom scanning. While the general concept is applicable to all satellites of this type, each satellite design has its own unique imaging system and requires unique methods to analyze the characteristic signatures. We assess the feasibility of detecting moving objects and measuring their velocities in one particular archive of satellite images made by Planet Labs Corporation with their constellation of SuperDove satellites. Planet Labs data presents a particular challenge in that the images in the archive are mosaics of individual exposures and therefore do not have unique time stamps. We explain how the timing information can be restored indirectly. Our results indicate that the movement of common transportation vehicles, airplanes, cars, and boats, can be detected and measured.
Paper Structure (9 sections, 3 equations, 8 figures)

This paper contains 9 sections, 3 equations, 8 figures.

Table of Contents

  1. Introduction
  2. Observations
  3. Characteristic signatures of moving objects
  4. Signature in composite visual images
  5. Signature in differenced images
  6. Millisecond timing from static images
  7. Measuring the velocity
  8. Altitude-speed ambiguity
  9. Conclusions and Future research

Figures (8)

  • Figure 1: Composite visual image from the Planet Labs data set 20220530_173806_19_241e. The image contains $13317 \times 9578$ pixels of $3\time 3$ m$^2$ covering $40.0 \times 29.7$ km$^2$ including the regions with no data.
  • Figure 2: Detail of the composite visual image (figure \ref{['visual-image']}) showing an airplane moving on or over the runway at the San Diego International airport. The image covers $1.13 \times 0.77$ km.
  • Figure 3: Detail of the composite visual image (figure \ref{['visual-image']}) showing boats moving in the San Diego harbor. The bright green leading edge is a characteristic signature of motion. The image covers $1.31 \times 0.78$ km.
  • Figure 4: Detail of the full image in eight spectral bands arranged in temporal order: top row, right-to-left, then bottom row, right-to-left. In this temporal order showing the airplane moving west (left), the spectral bands are B, R, G1, G2, Y, RE, NIR, CB (table \ref{['T2']}). Each panel covers $0.6 \times 0.6$ km.
  • Figure 5: Differences of the images in figure \ref{['8-band']} to isolate the moving object. The background is flatter if the images are subtracted in spectral order. Thus the 7 images show the differences CB-B, B-G1, G1-G2, G2-Y, Y-R, R-RE, RE-NIR (table \ref{['T2']}). The eighth square (black) has no data.
  • ...and 3 more figures