Abstract: According to the characteristics of WSN and the profile detecting system, a kind of methods of data processing based on WSN is proposed. Firstly, the method by principal component analysis extracted sample features, the features of different samples were then fused using accumulate mode and a mathematical model was proposed. On the basis of this model, a novel algorithm of special object recognition based on sparse representation in multiclass fusion sample was proposed by according to distribution of the main non-zero coefficient under a dictionary, and then the algorithm recognized the special target. Numerical simulation and experimental results demonstrate the effectiveness of the proposed algorithm, which means comprehensive performance was better than the traditional methods.

Key words: information processing , WSN, sparse representation, profiling recognition, unattended ground sensor