基于视觉认知机理的复杂动态背景下目标提取

A target extraction method in disordered and dynamic background based on visual cognition theory

针对现有复杂动态背景下目标检测与提取算法存在的局限性,提出构建一种基于视觉认知机理的时空域特征交互模式,以及基于该模式的复杂动态背景下的目标检测与提取算法。该算法通过时域特征在相似空域特征上的传递和空域特征依据时域特征一致性进行捆绑两个交互作用,实现复杂动态背景下的目标检测和提取。实验结果表明该算法不仅有效克服了传统目标检测与提取算法凭借单一特征或特征组合进行复杂背景中目标提取的不稳定性,同时大大降低了虚警率。

Aimed to the limitation of target detection and extraction method under clutter and dynamic background, a spatial-temporal features interaction model and the method for target detection and extraction were proposed, which was inspired the visual cognition theory of biology and based on the visual cognition theory. It was achieved by two parts that target detection and extraction under clutter and dynamic background. One is the transaction of temporal features through similar spatial features ; the other is bundle of spatial features according to the consistence of temporal features. Experiment results show that the proposed method can overcome the instability of traditional method which relay on signal ;eature of combination of features, at the same time the false alarm rate was decreased.