Using harvestable energy in Wireless Sensor Networks (WSNs) has gained considerable popularity recently. However, how to utilize the unstable power supply to achieve quality performance is still a challenge. In this paper, we propose a Weight-Optimized Source Rate Optimization (WOSRO) approach which allows the WSN system to adaptively control source coding rates by choosing the most valuable data packets to transmit. In this approach, the packet selection strategy is optimized by considering multimedia distortion reduction, energy cost, energy neutrality constraint and power saving efficiency. Simulation results show that the proposed data packet selecting strategy significantly improves data transmission quality by exploring harvesting-storage energy neutrality and multimedia data packet importance.