车载鱼眼图像的全景图拼接误差分析  

Error analysis of panorama stitching based on vehicle fisheye images

作　　者：曹君[1,2] 刘正华[3] 洪勇 邵慧超 

机构地区：[1]武汉城市职业学院,武汉430064 [2]立得空间信息技术股份有限公司,武汉430071 [3]中国地震局地震研究所地震大地测量重点实验室,武汉430071

出　　处：《测绘科学》2016年第7期202-204,172,共4页Science of Surveying and Mapping

基　　金：测绘地理信息公益性行业科研专项项目(201512008);科技部863项目(2015AA124001)

摘　　要：针对车载鱼眼图像拼接过程中的"拼接缝"的问题,对拼接误差的来源及其理论模型进行了分析。通过对全景影像的拼接过程进行梳理,对拼接误差的来源进行了论证,并对误差的理论模型进行了详细推导。最后,结合实际车载鱼眼数据,通过在不同目标距离情况下进行全景图拼接实验,证实了实际拼接成像效果与根据理论模型推导所得的拼接缝的大小基本吻合,说明了该文所提出的理论模型的可靠性,从而为后续改善全景图像拼接效果提供了理论支持。Aiming at the problem of ＂seam＂ which arising in the process of panorama stitching based on vehicle fisheye images, this paper analyzed the sources and theoretical model of the stitching error. Through elaborating the process of stitching panoramic image, the sources of stitching error were introduced and the theoretical model of stitching error was derived. The results of panorama stitching experiment at different distances with real vehicle fisheye data confirmed that the seam size of actual stitched panorama and that derived from the theoretical model were basically consistent, which showed that the model was reliable and could provide theoretical support for the follow-up work of improving the panoramic image mosaic effect.

关 键 词：车载 鱼眼图像 全景拼接 误差 

分 类 号：P208[天文地球—地图制图学与地理信息工程]