Using a mobile sink for data collection from the sensors has been considered as a good way to prolong the network lifetime ofwireless sensornetworks (WSNs). To avoid the problems of long delay and data non-fresh, some previous studies selected a set of data collection points (CPs) and all the other sensors transmit their data to the closest CP in a multi-hop manner. Then the mobile sink only needs to visit and collect data from the selected CPs, reducing the path length of the mobile sink and the time required for collecting data from all sensors. This paper proposes a data collection mechanism, called DDCF, which uses a mobile sink to collect data in a heterogeneous WSN(HWSN), aiming to prolong the network lifetime while improving the surveillance coverage. The proposed DDCF mainly consists of CP selection and tree topology construction phases in each round. The CP selection phase takes into consideration multiple parameters including remaining energy, coverage contribution as well as data fusion degree and then dynamically selects CPs in each round for balancing the lifetime of CPs and improving the surveillance quality. In the tree topology construction phase, each sensor selects its parent by considering the remaining energy, data fusion degree and transmission success ratio, which aims to dynamically construct a tree topology for further reducing and balancing the energy consumptions of sensor nodes. Performance study shows that the proposed DDCF outperforms existing studies in terms of network lifetime, fairness as well as surveillance quality.