Research of Multi-band Image Recognition Applied to Ships

Abstract

Aiming at the shortcomings of the traditional local feature recognition method of multi-band ship image target, the recognition speed is slow, and a local feature recognition method of multi-band ship image target is proposed. First, by detecting key points and feature matching, the local features of the multi-band ship image target are extracted; then, the multi-classifier combination is used to train and identify the local features through the linear classifier and the non-linear classifier, so that the design of local feature recognition method of multi-band ship image target is completed. Through comparison experiments, compared with the traditional local feature recognition method of multi-band ship image target, the experimental results show that the proposed local feature recognition method of multi-band ship image target has faster recognition speed.