In combination of the characteristic of JPEG2000 and the network architecture of Wireless Sensor Networks (WSNs), a multi-node cooperative JPEG2000 image compression method based on neighbor clusters is proposed by using the concept of in network processing to improve their energy,meomory and computional power. Firstly, the images captured by the camera node are partitioned into tiles and the adaptive bit rate is allocated by a optimum method based on gradient magnitudes. Then the tiles are sent to the neighbor clusters and compressed by multiple nodes cooperatively to minimize the total energy consumption of the network. Simulation results show that the proposed image compression method not only realizes the JPEG2000 compression of big size images in the WSNs, but also balances the energy load among all nodes. Compared with the compressing image before sending it to the base station, the proposed method has decreased the energy consumption of the camera node to 3.4%. These results show that the lifetime period of the network has improved by 7 times.