基于人工蚁群的红外弱小目标检测方法

提出一种基于人工蚁群的红外弱小目标检测方法.首先利用模糊隶属度函数对图像进行预处理获取可能的目标点;然后利用该结果与信息素一起共同指导蚂蚁的行为;最后通过蚂蚁行走路径上的信息素分布进行更新,使得分布在目标路径上的信息素逐渐增大,逐渐向真实的目标图像收敛达到抑制背景的目的.仿真和实验表明,与固定权值滤波方法和最大化中值滤波方法相比,该方法能够有效地从复杂背景的图像序列中检测弱小目标.

A measure method for infrared dim targets based on ant colony algorithm

An ant colony algorithm is presented to detect dim targets against the complex cluttered background in IR image. Firstly the image is decomposed by fuzzy membership function to obtain the possible targets. Then, the results are used as heuristic information, guiding ant colony with the pheromone together. By the distribution of pheromoae on the route that the ants passed updates, and the pheromones on the possible target points increases, the searching routes converge on targets image progressively based on the pheromone updating rule. Finally, the real targets by final intensity of the legacy of the pheromonese are extracted. Simulations and experiments show the infrared image can be detected from complicated background compared with the method based on constant weight coefficient filtering and max median filtering.