A collaborative probability based multimodel target identification method for the applications in wireless sen-sor networks was proposed.Wavelet package was used to extract the features of the target information sample

and in or-der to decrease the computing complexity and reserve the key information as well

feature compaction

which achieved by principal components analysis

was carried out before learning.Different types of Gaussian classifier were developed for classifier learning.To gain global optimal results

committee decision was employed to select and combine individual decisions with dynamically adjusted weights.The experiment results verified that the method introduced can adapt to the interference factor of the environment and can implement target identification accurately and stably.