Determining the Geographical Location of Image Scenes based on Object Shadow Lengths

摘要

Many studies have addressed various applicationsof geo-spatial image tagging such as image retrieval,image organisation and browsing. Geo-spatial imagetagging can be done manually or automatically with GPSenabled cameras that allow the current position of thephotographer to be incorporated into the meta-data of animage. However, current GPS-equipment needs certain timeto lock onto navigation satellites and these are therefore notsuitable for spontaneous photography. Moreover, GPS unitsare still costly, energy hungry and not common in mostdigital cameras on sale. This study explores the potential of,and limitations associated with, extracting geo-spatialinformation from the image contents. The elevation of thesun is estimated indirectly from the contents of imagecollections by measuring the relative length of objects andtheir shadows in image scenes. The observed sun elevationand the creation time of the image is input into a celestialmodel to estimate the approximate geographical location ofthe photographer. The strategy is demonstrated on a set ofmanually measured photographs.