一、论文简述1.第一作者:ZheZhang2.发表年份:20233.发表期刊:IJCNN4.关键词:MVS、深度学习、遮挡、双向投影5.探索动机:现有的工作很少考虑遮挡问题,导致边界和遮挡区域的重建效果不佳。IntraditionalMVSmethods,takingCOLMAPasanexample,theocclusionissuecanbemodeledundertheprobabilisticframework.However,veryfewlearning-basedmethodshavetakentheocclusionproblemintoconsideration.Howeve