Abstract: A target recognition system based on optical vision for AUV(autonomous underwater vehicle) is presented, and the hardware structure and software system of the vision system are described.Underwater image processing techniques are introduced,and considering the influence of scale factors on moment invariant features in discrete conditions,affine transformation invariants based on invariant moments are constructured.Structure and modeling of the immune genetic neural network(IGNN) applied to target recognition are discussed in detail.Experiments of underwater target recognition are made on four types of actual targets,and the results prove that the presented underwater target recognition system can be applied to underwater image recognition with relatively high real-time performance and accuracy.