A Human-Attention Driven Video Capture Method for Wearable Visual Computing
In this paper, we propose a human-attention driven video capture method for wearable visual computing. This method could help wearable vision system to avoid collecting a lot of meaningless visual images and save the platform resources. We firstly calculate background similarity and recognize user’s attention behavior, then detect user’s active or passive attention status, finally automatically and selectively capture video information according to user’s attention interest. We describe architecture of proposed method and give an experiment to validate its feasible.
L. Han et al., "A Human-Attention Driven Video Capture Method for Wearable Visual Computing", Key Engineering Materials, Vols. 467-469, pp. 1361-1366, 2011