移动视点下室外场景的实时阴影检测

Real-Time Shadow Detection from Live Outdoor Scenes with Moving Viewpoints

提出一种面向增强现实应用的框架,用于实时检测在移动视点下室外场景中的阴影。该框架将每帧分为光流跟踪得到的跟踪区域和被新引入场景中的新进区域。跟踪区域中,该框架利用光流计算当前帧的阴影边缘,同时使用阴影边缘sigmoid模型对光流累计误差进行校正;随后对边缘提取特征,为新进区域区分阴影和非阴影边缘,并且加以空间布局约束以消除伪影。实验结果表明,该框架可以准确地检测和跟踪移动视点下的阴影,并应用于增强现实中。

A framework for augmented reality applications is proposed for real-time detection of shadows in outdoor scenes from a moving viewpoint.This framework divides each frame into a tracking region obtained by optical flow analysis and a emerging region newly introduced into the scene.In the tracking region,the framework uses optical flow to calculate the shadow edges of the current frames,and at the same time uses the shadow edge sigmoid model to correct the cumulative error of the optical flow;the features of edges are then extracted to distinguish the shadow and non-shadow edges for the emerging region,and spatial layout constraints are further considered for the emerging region to remove spurious shadows.Experimental results show that the framework can accurately detect and track shadows under moving viewpoints and apply it to augmented reality applications.