Multi-spectral misregistration of Sentinel-2A images: analysis and implications for potential applications

Area of Science:

• Remote Sensing
• Earth Observation
• Image Analysis

Background:

• Sentinel-2A Multi-Spectral Instrument (MSI) images have inherent sub-pixel shifts due to detector parallax.
• Existing pre-processing routines correct for static shifts but not for moving or high-altitude objects.
• Sub-pixel misregistration analysis offers a novel approach for detecting dynamic phenomena.

Purpose of the Study:

• To analyze sub-pixel misregistration in Sentinel-2A MSI images.
• To explore the potential of this misregistration for detecting moving targets and clouds.
• To evaluate the effectiveness of phase correlation for identifying shifts.

Main Methods:

• Phase correlation approach applied to Sentinel-2A MSI bands B2 (blue), B3 (green), and B4 (red).
• Detection and quantification of sub-pixel shifts between image acquisitions.
• Comparison of detected clouds with the Sentinel-2A Level-1C product's built-in cloud mask.

Main Results:

• Observed shifts exceeding 1.1 pixels for moving targets, including airplanes and clouds.
• Demonstrated capability of the phase correlation method for cloud detection.
• Identified clouds missed by the standard Sentinel-2A Level-1C cloud mask.

Conclusions:

• Sub-pixel misregistration analysis is a viable technique for detecting moving targets in remote sensing data.
• This approach improves cloud detection accuracy, identifying features missed by conventional methods.
• Phase correlation offers a valuable tool for enhancing the analysis of Sentinel-2A imagery.