摘要
随着无人机技术的发展,利用无人机影像进行地形三维重建和全景影像生成,已经越来越普及。然而由于无人机影像拍摄时角度、光照条件、相机设置等具有不一致性,多视角无人机影像间通常存在一定的颜色差异。颜色差异的存在,会影响多视影像密集匹配的精度,从而影响三维几何模型的精度。此外,颜色差异也会影响全景影像和纹理模型的视觉效果,拼接后影像及纹理模型上易出现“鬼影”及颜色拼接缝等问题。为解决该问题,本文提出一种无人机多视影像颜色一致性处理方法。首先,采用Structure from Motion (SfM)方法获取无人机影像间的所有匹配点;其次,构建并简化多视角无人机影像拓扑关系图;最后,基于颜色对应关系全局优化所有影像的颜色映射模型,获得匀色后影像。通过真实数据集进行实验,验证了本文提出的色彩一致性处理方法的有效性。
With the development of UAV technology,the use of UAV images for terrain 3D reconstruction and panoramic image generation has become increasingly popular.However,due to the inconsistency of angle,lighting conditions,camera settings,etc.when UAV images are taken,there is usually a certain color difference between multi view UAV images.The existence of color difference will affect the precision of dense matching of multi view images,thus affecting the precision of 3D geometric modeling.In addition,color difference will also affect the visual effect of panoramic images and texture models.After stitching,images and texture models are prone to "ghost" and color stitching.To address this issue,this article proposes a color consistency processing method for unmanned aerial vehicle multi view images.Firstly,the Structure from Motion(Sf M) method is used to obtain all matching points between drone images;Secondly,construct and simplify the topology relationship diagram of multi view drone images;Finally,based on the color correspondence relationship,global optimization the color mapping model of all images to obtain the uniform color image.Experiments were conducted on real datasets to verify the effectiveness of the color consistency processing method proposed in this paper.
作者
张哲
雷帆
蒋琦
杨凯钧
Zhang Zhe;Lei Fan;Jiang Qi;Yang Kaijun(Key Laboratory of Natural Resources Monitoring and Supervision in Southern China Hilly Region,Ministry of Natural Resources,Changsha Hunan 410119)
出处
《国土资源导刊》
2023年第2期49-54,共6页
Land & Resources Herald
关键词
无人机影像
色彩一致性处理
匀色
多视角影像
UAV imaging
color consistency processing
uniform color
multi perspective imaging
