Exploring the Novice Approach to Orthorectification of Satellite Imagery

Abstract

Orthorectification plays vital role in satellite image processing. This process imposes challenges due to the dynamism in capturing environment, capturing unit, satellite rotation, sensors parameters and overlying regions on earth. The required geometric modelling needs an accurate estimation of Ground Control Points (GCPs) and their processing. Most of the proposed models are computational intensive and use manual approach for locating GCPs. Further, GCPs co-ordinates are floating point numbers the computational capability of the system imposes the constraint on the accuracy and robustness of the respective models. In this paper we have studied orthorectification process and proposed instinctive processing framework for orthorectification of optical pushbroom sensor based satellite imagery. The frame work accompany metadata extraction, automatic ground control point (GCP) extraction using parallel processing, geometric modeling, orthorectification and image stitching processes. Experimental results with proposed framework confirmed the robustness of the technique and provided sub pixel accuracy on independent check points with positional accuracy around one pixel for orthoimage. Parallel SIFT features are extracted using SIMD architecture while performing image stitching.