With the widespread popularity of sensor-rich mobile devices, such as Android phones and iPhones, crowdsensing networks have attracted enormous attentions from researchers recently. Mobile devices, carried by people in their daily life, move randomly in monitoring areas and acquire environmental sensing data. However, it is an arduous task to design an efficient data gathering approach for crowdsensing since the movements of mobile devices are difficult to predict. In this paper, a location based data gathering scheme for multi-sink crowdsening networks is proposed. First, a selection method based on multi-objective decision-making is presented to find the optimal sink according to the distances, connection durations and encounter probabilities between mobile devices and sinks. Then, inspired by PeopleRank algorithm, a location based sensing data forwarding algorithm is presented to optimize the data forwarding strategy. Furthermore, the feasibility and effectiveness of the proposed scheme are validated through a series of experiments employing opportunistic network environment simulator. The results show that the proposed scheme not only improves the transmission rate of perception data but also has lower cost and forwarding delay than the existing methods.