Feature based vessels classifications from partial view image

Abstract

Automatic object recognition has diverse applications in various fields of science and technology ranging from military to civilian industries. It can provide better tracking and automatic monitoring to control from potential enemy ships. Classification of objects based on their silhouettes is particularly useful in autonomous ship recognition. However, problem arises when part of object becomes invisible (e.g. due to partially shifting out of view) or partially occluded by with another silhouette; see Figure 1 for an example. For clipping conditions, the shift level indicates the fraction of columns by which the object has been shifted to the right. While for occlusion, the overlap level indicates the fraction of columns which have been corrupted by the secondary silhouette, which may be stationary. In this case, when moving detection algorithm is involved , only parts of the moving vessel will appear.