海面红外目标智能跟踪算法研究  

INTELLIGENT TRACKING ALGORITHM FOR INFRARED OBJECT ON THE SEA

作　　者：王新赛[1] 张天序[1] 

机构地区：[1]华中科技大学图像识别与人工智能研究所 图像信息处理与智能控制教育部重点实验室,武汉430074

出　　处：《模式识别与人工智能》2002年第1期66-69,共4页Pattern Recognition and Artificial Intelligence

基　　金：国家自然科学基金(69875005)

摘　　要：根据海面运动目标及成像传感器的特点,本文提出了一种红外目标图像序列跟踪点自动识别的智能跟踪算法.以平均梯度为特征进行目标图像分割;并根据目标最亮区与吃水线的空间关系选择目标最亮区中心点的垂线与吃水线的交点作为跟踪点;利用D-S证据理论融合帧间关联特征的跟踪点的自动识别以及可信度确认.实验证明本方法合理而且十分有效.In accordance with the characteristics of moving objects on the sea and of the imaging sensor, an intelligent tracking approach about automatic recognition of the tracking-point in infrared object images is presented in this paper. The algorithm of image segmentation is based on the feature of gradient images. And on the basis of the spacial relationship between the brightest area and waterline of the object, the intersection of the waterline and the vertical center line of the brightest area is selected as the tracking point. Making use of D-S' s evidence theory, the method of fusion correlation features between frames is used to automatically recognize the tracking-point and affirm reliability. The experiment testified that the method is reasonable and very effciency.

关 键 词：梯度特征 目标分割 证据理论 可信度 图像序列 自动识别 海面红外目标智能跟踪算法 

分 类 号：TN215[电子电信—物理电子学]