Invention Grant
US08358380B2 Efficient spatial and temporal transform-based video preprocessing 有权
高效的时空变换视频预处理

Efficient spatial and temporal transform-based video preprocessing
Abstract:
In one method embodiment, receiving a first temporal sequence of video frames, the first temporal sequence corrupted with noise; frame matching the video frames according to a first stage of processing; denoising the matched frames according to a second stage of processing, the second stage of processing commencing responsive to completion of the first stage of processing for all of the video frames, the second stage of processing comprising overlapped block processing; and wherein denoising further comprises accumulating denoised pixels for each iteration of the overlapped block processing in a two-dimensional (2D)+c accumulation buffer, the 2D accumulation buffer corresponding to the denoised pixels corresponding to a reference frame of the video frames, where c comprises an integer number of non-reference frame buffers greater than or equal to zero.
Public/Granted literature
Information query
Patent Agency Ranking
0/0