Region of Interest Coding in Photsat Mission

The Institute of Space Studies of Catalonia (IEEC) coordinates the PhotSat project [1], which involves the development and construction of a satellite to monitor the 10 million brightest stars in the sky for at least two years. The scientific objective of this project is to complement the Large Synoptic Survey Telescope (LSST) [2] and the GAIA mission [3] by examining the intensity of stars. In the PhotSat mission, a CubeSat is expected to be launched at the end of 2025, with the mission of scanning the entire sky using two cameras—one with an ultraviolet detector and another with a visible detector. The desired configurations of the camera field of view (FoV) and the sensor step size are 8◦ and 4◦, respectively, which will generate a total of 16.75 GB of image data that will need to be transmitted from the satellite to the Montsec ground station [4]. Since the download capacity to the Montsec station is limited to 8 GB per day, some compression is required.

The compression presented in this paper consists of two stages. The first stage uses an auto- matic object detection technique based on photometry [5], which analyzes the pixel intensities of the acquired data to identify the pixels corresponding to the Regions of Interest (ROI). This pro- cess adjusts the image so that only the ROIs are preserved, eliminating irrelevant pixels. In the second stage, the modified data is losslessly compressed using a prediction-based compressor and a contextual binary arithmetic encoder [6]. To provide contextual information around the ROI, the ROI can be expanded using the parameter ROI Extended Coding Pixels (RECP), which adds a specified number of pixels around the detected objects. Additionally, both the predictor and ROI coding can be independently enabled or disabled, allowing for faster execution and, under certain circumstances, improved coding performance.

Experimental results show that when our proposal is used with ROI coding disabled, all acquired data can be losslessly compressed to 5.86 GB. However, when ROI coding is enabled, compression performance improves significantly; for example, if RECP is set to 0 or 64, the acquired data can be compressed to 0.16 GB and 4.13 GB, respectively—a very significant compression.