Scatterable landmines are a recent warfare technology that can be dispersed aerially and quickly render land useless to local populations. This research aims to develop methods for civilian landmine detection and suspected area reduction using readily available unmanned aerial systems (UAS), simulation landmines and very high spatial resolution orthomosaics derived using Structure from Motion from multiple flight dates. This study demonstrated image processing for uniform target detection combined with a multi-temporal analysis, has potential for reducing the search area for landmine detection and making the work of civilian deminers significantly safer. Results indicate it is possible to detect individual landmines, reduce the area needed to search to less than three percent of the original study area, and a multi-temporal process can reduce the number of false positives detected. The best practices for data collection and analysis are reported, and possibilities for future demining methodologies are suggested.
