Considering the fact that the wireless sensor networks (WSNs) need to maintain a long lifetime, there is a great demand to decrease energy dissipation of the sensor. Data compression is an efficient method to solve the problem. This paper proposes a practical and efficient data compression algorithm with high compression and noise-resisted features, in which the quasi-cyclic low-density parity-check (QC-LDPC) codes and the Kalman filters are used to compress the transition data of the sensors and to provide the side information for the joint decoding, respectively. The simulation results prove that the algorithm provides an outstanding performance than the famous syndrome techniques.