Research Of Automatic Vessel Target Recognition Technology In Imaging-typed Guidance System

The research background of this paper is automatic vessel target recognition in imaging-typed guidance system. We primarily realized the ship recognition function for infrared images and color video images in algorithm and software aspect.First of all, we need a three-dimensional ship model database for training classifier. Five kinds of ship were chosen to be included in the recognition class and then be modeled: an aircraft carrier, a frigate, a destroyer, a research ship, and a merchant ship.Second automatic vessel recognition work can be divided into three steps. The first step segments the infrared images and color-video images to obtain the ship silhouette. The second step calculates 6 invariant moments of region, 6 invariant moments of boundary and one normalized moment of inertia feature of those silhouettes, using these thirteen features as basic pattern. The third step directly classifies the ship as one of the five types with k-nearest neighbor method.We used 25 FLIR images and 15 color video images to test our ship recognition system. The number of test image is too limited to estimate the system performance, but there are many aspects to be improved in the system through this testing.