Towards better propagation of geographic location in digital photo collections

Abstract

The integration of GPS in smartphones, tablets and digital cameras become increasingly popular, but GPS receivers does not work well indoors. This malfunction can generate erroneous location information from where the picture was really taken, or no information at all. In order to overcome this problem, this paper proposes an automatic selection of techniques that couples different solutions to the problem from multiple regression. The focus of the work is to minimize the error generated by existing techniques with an automatic selection of techniques that uses Artificial Intelligence methods to automatically select the best technique. Also, we execute an experiment to validate the results achieved. After validation, the tests indicated that the proposed automatic selection technique improved the results in five of the six scenarios. In the sixth scenario the results were the same.