Automatic Graben Detection in Lunar Images Using Hessian Technique

Lunar surface exploration is increasing rapidly with priority being given to precision of lunar soft landing. Lunar soft landing is achieved when craters and grabens are used as navigational landmarks. Grabens are formed from localized tensional stress fields or from near-surface dike emplacement. These tectonic features tend to have consistently straight or accurate parallel-striking walls bounded by steep, inward-dipping normal faults. Aiming at navigational application, a novel approach for automatic graben detection based on Hessian technique has been implemented on Digital Terrain Model (DTM) of lunar image. The Hessian technique uses gradient change as a key parameter to identify grabens. Adaptive Binarization using Otsu method is used to extract graben features from the Hessian image. Features such as small grabens and craters are removed using morphological operations, resulting in significant appearance of grabens. The experiment is conducted in different DTM images of lunar surface and the results indicate 90 % of the grabens are detected. The statistical results are evaluated based on visual interpretation, for both automatic and manual graben detection. It is observed that the proposed automatic graben detection technique gives better results than the manual detection.