Geographic profiling uses locations of series crimes to detect the most probable residence of the offender. However, because of shortage of other information, the method doesn't perform well. To Solve this problem, the concept of geographic profiling was extended and a concept of human geographic profiling (HGP) was put forward to take offender’s bias, victims’ characteristics, and geographical circumstances into account. The physical geographic profiling (PGP) was obtained by combining an improved “spatial mean” method and an improved criminal geographic targeting (CGT) model by using a precisely deduced "Combination Function". Afterwards, HGP and PCG were combined by the same function. The resultant product was a three-dimentional probability map. Simulation result indicated high accuracy rates among which the lowest is 66% under an ideal situation. The introduction of HGP and deduction of combination function are discussed originally and are of great importance in this paper.