The detection of geological faults in high resolution satellite and aerial photos is of great interest in several tasks such as cartography, underground or superficial characterization, the location of spring of water, the estimation of the stability of the slopes, and the driving in the extraction of stones. This duty is usually based on the identification of certain geometrical features through campaigns and handmade analysis of the photos by using expensive hardware and software. Assuming a fault as the union of consecutive dark segments, we have implemented a variety of algorithms to highlight such structures with different degrees of details. The efficiency of the detector has been verified on a database of real images, showing that it is possible to analyze complex structures in a very fast and unsupervised way.

Genetic discrete reconstruction.