针对潜望镜的舰船目标识别算法的研究

Research on The Algorithm of The Ship Target Recognition Using Periscope

鉴于潜望镜智能化程度不高的现状,对以潜望镜识别水面舰船为应用背景的目标识别算法进行了研究。算法选择Zernike矩不变量作为识别特征,并且利用基于上层建筑的矩不变量以提高特征的分类能力;同时还对舰船目标的观察空间进行形态划分,减少目标表示所需图像数量,提高识别效率。仿真试验验证了识别算法的有效性。

In view of the low intellectualizer of periscope,the recognition algorithm is researched in this paper, which is based on the background of the ship target recognition using periscope. Zemike moment invariants are selected as the recognition features in this algorithm, and are computed based on the superstructure in order to improve the discrimination ahility. The horizontal viewing sphere of ship target is divided into several aspects so as to reduce the amount of descriptive images and improve the recognition efficiency. The experiment shows that the recognition algorithm is very valid.