Abstract: In our paper

we introduce the basic principle and algorithm of the Complex Valued ICA

and a new target recognition algorithm based on Complex Valued ICA is proposed. With our method

we must compose complex value training group matrix with two images from different cameras

secondly

the Fast ICA algorithm is performed on the matrix to get independent components (ICs)

in the third step

we need extract the training set and testing set’s features based on ICs

then Fisher discriminant analysis is introduced to train target features’ pattern

at last

the classification of targets in testing set is performed. Our experimental data show that the method proposed in our paper is feasible and the target recognition rate is satisfied.