Most of the existing state-of-the-art studies assume that there is only one task requester for the crowdsourcing systems, which is hard to satisfy the scenario of multiple task requesters. Thus, this work focuses on the crowdsourcing systems consisting of multiple task requesters, where all the requesters and workers can bid or complete multiple tasks in each round. On the basis of that, this paper proposes a truthful double auction mechanism for crowdsourcing systems to incentivize workers and allow efficient task completion. First of all, this method abstracts the workers that interested in multiple tasks into multiple virtual workers, and further designs an efficient task assignment mechanism based on the principle of low bids first matching. A truthful pricing mechanism is developed based on the principle of uniform pricing. In addition, a proof is given to show that the proposed mechanism satisfies three essential economic characteristics:truthfulness, individual rationality, and budget balance through substantial theoretical analysis. Finally, extensive simulations are conducted to study the performance of the proposed auction mechanism, and the simulation results corroborate the theoretical analysis.