基于图像拼接的全景目标检测技术  

Panorama Target Detection Techniques Based on Image Mosaic

作　　者：陆天舒[1] 毛征[1] 郑利苹[1] 彭超[1] 

机构地区：[1]北京工业大学电子信息与控制工程学院,北京100124

出　　处：《兵工自动化》2014年第2期7-10,共4页Ordnance Industry Automation

摘　　要：针对战场指挥车等战地车辆内作战人员需要在车内实时了解外部战场动向的问题,提出一种基于图像拼接的全景目标检测技术研究。利用监控转台摄像头获得全景图像并检测运动物体,以此模拟战场实际情况。首先使转台进行水平360?旋转,同时使用安装在转台上的摄像头采集实时图像,利用转台转速与扫描场时间等参数对图像进行预处理,从而削弱前后两场信号叠加造成的边缘模糊。最后对前后两帧图像进行灰度值相关匹配,并检测出全景图像上的运动物体。实验结果表明,该技术可以实现全景图像拼接并在全景图中标出视场内运动物体的位置。Aiming at the battlefield command vehicle operators needing to know the outdoor battlefield situations in real time, put forwards the panorama target detection technology based on image mosaic technology. Use monitoring rotation platform camera to acquire panorama image and detect motive object, and simulate battlefield practical situation. At first, the rotation platform rotates 360° in horizontal orientation, at the seam time use camera to acquire real time image, then carry out image pretreatment based on parameters such as rotation platform rotating speed and scanning field time. Therefore the fuzzy edge cased by signal overlay is reduced. At last, carry out gray level correlation matching of two images and detect the motive objects on panorama image. The test results show that the technology can realize panorama image mosaic and label motive object location in panorama image.

关 键 词：灰度相关 图像拼接 光电目标检测 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]