This paper addresses the localization embedded in a wild environment with huge obstacles and limitations of energy and costs. This kind of wild large scale localization often selects the range-free approaches for the reason of hardware consumption. While these approaches always suffered from low accuracy compared with the range-based methods, especially in the scenario of wild environment with huge obstacles and hardware limitation. Here we introduced a regional partition and cooperation (RPC) localization method and applied it to sensor localization for the Great Wall site in Yulin, Shaanxi province. We proposed a novel regional partition method to divide the large scale scene into small sub-regions which have no obstacles, consequently the well known DV-Hop is used for sub-region location, in the end a new sub-region cooperation method is provided to increase the accuracy, it should be noted our regional partition method and sub-region cooperation method only use the count information and sensor connectivity information. Theory analysis shows the proposed RPC can locate all unknown sensor nodes only requires the information of hop count and connectivity, furthermore RPC works even the anchor nodes are deployed sparsely, meanwhile the sensor nodes do not need any extra distance-measuring equipment. In the simulations, the positioning error of per node, the performance versus different number of anchor node and different node density are analyzed, the results demonstrate the performance extensions compared with classic DV-Hop.