Electromagnetic wave velocity is the most crucial parameter in processing high frequency electromagnetic prospecting (ground penetrating radar: GPR) data. Migration techniques that strongly depend on velocity parameter have been used to refocus the scattered signals from the x-t domain or image space back to their true spatial location in the object space. To optimize migration algorithms, the authors propose using entropy as a criterion of looking for the correct propagation velocity. The goal of this article is to present the theoretical basis, the procedure as well as the necessity for combining migration with entropy. Using mathematical models, the authors demonstrated the need of having the migrated sections and the similarity between the estimated velocities and the root mean square ones up to the top of objects. The results verified on field data show that migrated images with true velocity have the best quality and minimum entropy, thus the size as well as the depth of anomalies is determined with high reliability.