UAV侦察图像自适应预处理仿真研究  

Simulation of UAV Surveillance Image Adaptive Preprocessing

作　　者：刘慧霞[1] 张清[2] 席庆彪[1,2] 

机构地区：[1]西北工业大学第365研究所,陕西西安710065 [2]西北工业大学自动化学院,陕西西安710072

出　　处：《计算机仿真》2012年第7期105-109,共5页Computer Simulation

基　　金：国家自然科学基金(61074155)

摘　　要：研究无人机(Unmanned Aerial Vehicle,UAV)侦察目标识别中的图像增强问题。针对UAV侦察活动中因云雾干扰和由于成像距离较远而造成的图像不清晰这两个常见的原因,提出一种自适应的预处理算法。首先判断侦察图像上是否存在云雾,若不存在,则采用指示滤波器(Guided filter)对图像进行增强,可以细化图像的边缘,增强图像的对比度,并能增加可用于识别的特征点;若存在云雾,则采用基于暗通道优先的图像去雾技术,对于侦察获取的薄雾图像可以较清晰的还原出图像的无雾场景,而在浓雾条件下,可以清楚的分辨出图像中的道路、村庄等有意义的目标。仿真结果证明,通过自适应预处理算法后的图像清晰度有所增强,可满足识别需求。Research image enhancement of UAV automatic target recognition. This article focused on two prob- lems which are common in UAV reconnaissance, one is weather interference, the other is image blur caused by high -altitude flight, and then proposed an adaptive preprocessing algorithm. Firstly, the existence of clouds in the image was determined, and if not present, the image was enhanced using guided filter which can refine the edge and in- crease the feature points of the image. If there are clouds in the image, then the image was dehazed using a dark- channel prior algorithm. In mist condition, this algorithm can restore the image scenes without fog. In fog condition, this algorithm can clearly distinguish the image of the road, the village and other meaningful goals. Simulation results show that, the image has been enhanced after the preproeessing and can meet identification needs.

关 键 词：预处理 无人机 图像增强 图像去雾 指示滤波器 暗通道优先 

分 类 号：TP317.4[自动化与计算机技术—计算机软件与理论]