We propose a multi-hop image transmission (MHIT) mechanism for the heterogeneous wireless multimedia sensor network (WMSN) composed of common nodes and image nodes, where the images are compressed by the nodes in the nodes’ neighborhood. According to the hop count and distance, the image node determines whether or not the image should be compressed before transmission. If it costs more energy to compress the image, the image will be sent directly. Otherwise, the task of image compression is distributed to the nodes in the neighborhood, which can balance the network energy consumption and help ease the energy burden on the image nodes. Experimental results show that the proposed mechanism can effectively solve the problem of energy hole caused by the image compression in the WMSN, and significantly prolong the network lifetime. So it is well suited for remote image transmission in large-scale WMSN.