Wireless sensor networks usually have limited resources

such as energy

bandwidth and processing and so on.And they can’t match the transmission of a large number of data.So

it is necessary to perform in-network compression of the raw data sampled by sensors.The data sensor node collected normally have multiple-modalities pertinence.Multiple-modalities pertinence refers to the different types of data which the same node sampled have some correlation.A adaptive multiple-modalities data compression algorithm using wavelet was designed.In a given threshold of the correlation

the data can be adaptive classified using this algorithm.the relevant data can be estimated using the least square estimation.The characteristics data are abstracted as a matrix

then can be exploited the spatial and temporal corrections using wavelet transform.Theoretically and experimentally

the proposed algorithm can effectively exploit the correlation of the data

the compression ratio of the algorithm has improved.Effectively

it can provide a significant reduction in energy consumption.