In this paper, a novel distributed algorithm is proposed to recover two-dimensional graph signals. This work exploits a Cartesian product graph as a graph model to present complex datasets. Furthermore, incorporating the piecewise smooth assumption, a LASSO optimization formulation is tailored to the recovery problem. To solve the optimization formulation in a distributed manner, the Cartesian product graph is decomposed into a family of overlapping subgraphs, and then a series of localized subproblems resided on the subgraphs are introduced. The subproblems are solved via an alternating direction method of multipliers (ADMM). Experiments on three real-world datasets demonstrate the excellent performance of the proposed algorithm.