Detecting Impact Craters in Planetary Images Using Machine Learning

Abstract

Prompted by crater counts as the only available tool for measuring remotely the relative ages of geologic formations on planets, advances in remote sensing have produced a very large database of high resolution planetary images, opening up an opportunity to survey much more numerous small craters improving the spatial and temporal resolution of stratigraphy. Automating the process of crater detection is key to generate comprehensive surveys of smaller craters. Here we discuss two supervised machine learning techniques for crater detection algorithms (CDA): identification of craters from digital elevation models (also known as range images), and identification of craters from panchromatic images. We present applications of both techniques and demonstrate how such automated analysis has produced new knowledge about planet Mars.

Characterisation of potential landing sites for the European Space Agency's Lunar Lander project

“…Branching Factor = FP/TP The Detection Percentage (correctly detected features over total number) is comparable with state of the art performances for craters (Stepinski et al, 2012, Kim et al, 2005. The Branching Factor instead indicates that, for every 10 true craters, 3 are false alarms, and that for each boulder almost 9 false identifications are counted.…”

“…The second processing step classifies the potential features as crater or boulder by exploiting their characteristic bright/dark pattern. This is done by means of image correlation between the respective pixel cluster and templates representing crater and boulder illumination pattern (Stepinski et al, 2012, Kim et al, 2005). The algorithmic concepts described above have been implemented in the MATLAB environment (MATLAB, 2010) and, whilst not new in essence, their implementation and the parameter settings have been tuned to meet the extreme illumination conditions in the lunar polar region.…”

“…True Positives (TP) are features (craters or boulders) that are detected and correctly classified; False Negatives (FN) are features that have not been detected; False Positive (FP) are features that have either been incorrectly detected (for example a non-feature undulation of the terrain) or correctly detected but incorrectly classified (for example a crater classified as a boulder or vice-versa). The following measures are also used: Detection Percentage = 100•TP/(TP+FN) Branching Factor = FP/TP The Detection Percentage (correctly detected features over total number) is comparable with state of the art performances for craters (Stepinski et al, 2012, Kim et al, 2005. The Branching Factor instead indicates that, for every 10 true craters, 3 are false alarms, and that for each boulder almost 9 false identifications are counted.…”

Detection of Small Impact Craters via Semantic Segmenting Lunar Point Clouds Using Deep Learning Network

“…A range of studies [6][7][8]14] compared numerous machine learning methods. However, the mentioned machine learning methods are now being replaced by deep learning methods.…”

“…Crater detection algorithms (CDAs) aim to perform crater detection more efficiently and robustly. Among many surveys [6][7][8], the CDAs can be divided in a range of manners. The most common is to divide CDAs into non-machine learning methods and machine learning methods.…”