Genetic sun shadow positioning model for digital forensics

Abstract

Extensively growing a set of methodologies to collect, provide, and analyze geoinformation in this decade. While due to the privacy concerns or elevated consideration, only less than half of videos and pictures bring geographical tags. Meanwhile, such geographical indications are not always available and reliable because of device dependencies. For example, users forbid location functions on devices or manipulate published geo-tags. In this case, the straightforward images' data for digital forensics will lose value. In this thesis, we propose an approach for shadow positioning model to smooth over this obstacle, such as non-geo-tagged information in videos and photos. Firstly, we briefly summarize existing digital forensics researchers rely on geotagging and figure out the potential limitations and threats for compromising the geo-information from videos and photos; then present a positioning model and algorithms are based on Solar Position Algorithms (SPA) and Genetic Algorithm (GA) to estimate geo-coordinates for non-geo-tagged source files. The experimental results show that our proposed model and algorithm can successfully compute a set of latitude and longitude values, and the average error in ±0.2° for both latitude and longitude. These combined algorithms derive from the theory of astronomy and evolution, which shapes a novel way to obtain geolocation information from such consequences.