© 2016 The Authors. This paper presents initial results of a JavaTM based, feature extraction tool, which represents a standard implementation of a hill-shading algorithm that transforms a 2D image to pseudo 3D image to enhance edge contrast in combination with an edge detection Canny algorithm that performs segmentation to produce multidirectional sun-shaded images and their edges. Our goal is to firstly automate this processes in JavaTM to obtain multidirectional optimization of edge discovery and secondly scale this algorithm to the complete SRTM raster collection at multiple pixel resolutions to document the distribution of Earth topographic discontinuities from continental to regional and local scales, respectively on the order of 1000s, 100s and 10s of kilometers. This tool will support the automatic extraction of lineaments of the transformed images to predict the existence of linear features that can be often found in association with ore deposits and landslides, if they represent tectonic lineaments. The collection of processed big data, represents a multi-scale data repository that may find use for these and other geological and environmental applications. We present preliminary outputs from a case study conducted in the Flin-Flon greenstone belt in Canada, which is well known for its base-metal endowment. In this study, two main shaded relief images with multidirectional illumination were created in Java each with four azimuth angles of the light sources and from which our developed tool extracts automatically multiple lineaments. The extracted lineaments represent both positive and negative elevation breaks, due to sudden slope inversions identifying dominantly crest lines and valleys. Preliminary results show good agreement with drainage networks, mapped fault lines and orientations of structures measured in the field. The main trends of the extracted lineaments of both images are NW-SE, N-S, E-W and NE-SW.