Hybrid method for detection of lunar craters based on topography reconstruction from optical images

Many applications in lunar and planetary science require robust crater detection algorithms (CDAs). In this paper, we present a novel hybrid CDA which utilizes the topography reconstructed from optical images and our previously developed topography-based CDA. The proposed CDA is applied to two Chandrayaan-1 M³ optical images and one selected region of Lunar Reconnaissance Orbiter WAC global optical image mosaic. The overall approach consists of: (1) reconstruction of topography from optical images; (2) corrections of brightness and contrast of used optical images (this step is required for manual verification of crater-candidates before inclusion into the resulting catalogue); and (3) processing by topography-based CDA. In addition, the integrated topography-based CDA has been used for the review of the initial topography reconstruction algorithm, and consecutive improvements. With such an approach, we demonstrate that our topography-based CDA, additionally improved using new crater shape-based interpolation method, can be used: (1) with topography refined from optical images; and (2) during subsequent improvements of the topography reconstruction algorithm. Experimental evaluation of the proposed CDA is done by manual verification of crater-candidates and registration in the previous LU60645GT catalogue. The evaluation has shown that the proposed CDA was used successfully for cataloguing of 3570 Lunar craters, which are almost all visible craters from the three selected regions. The accompanying result is the new LU64215GT catalogue of 64 215 Lunar craters, which is currently the most complete catalogue of Lunar craters.