Object Detection of Jupiter's Features Through NASA's JunoCam: an Eigenimaging Approach

Abstract

The JunoCam is a moderate-resolution camera mounted on the Juno spacecraft, launched by the National Aeronautic and Space Administration (NASA) in 2011, used to generate images of Jupiter of a previously unseen quality for both research and public outreach. Detailed Red, Green, and Blue (RGB) images from JunoCam can be used in conjunction with object detection algorithms to track visual features of Jupiter’s atmosphere. The positioning and movement of these features, such as circumpolar cyclones, hazes, and more, can provide insight into the nature of Jupiter’s atmosphere. This positional data can be used in conjunction with infrared and microwave data from Jovian InfraRed Auroral Mapper (JIRAM) and Microwave Radiometer (MWR) to give a more comprehensive assessment of Jupiter’s atmospheric qualities. The object detection method to be used in this research is eigenimaging, an approach adapted from the facial recognition method of eigenfaces. This methodology uses the eigenvectors of images projected onto a known eigenspace to match images from a new perijove capture to those in a dataset of previous perijoves. The efficient detection and classification of features across perijove captures is the ultimate goal of this method, wherein further analysis will extract the desired positional data. The objective of this proposed research is to prove that eigenimaging is a viable tracking method for white storms of various sizes on Jupiter's surface and to provide a framework for expanding the algorithms tracking mechanisms and trackable features.