Spatio-temporal noise filter for video encoders.
Publication Date: 2002-Apr-18
The IP.com Prior Art Database
Spatio-temporal noise filter for video encoders
Purpose: To reduce noise in a video sequence in order to improve video quality and
decrease bit rate.
General description: The video noise filter considered in this work is an adaptive
averaging spatio-temporal IIR filter proposed by L. Camiciotti and F. Bruls . This noise
filter uses two past filtered frames to filter the current frame, thus accomplishing spatiotemporal
noise suppression. Typical video encoder designs operate on macro-block (MB) data
. This necessitates a MB-base noise filtering.
MB-based noise filter: Fig. 1 depicts the adaptive averaging spatio-temporal filtering
operation to suppress noise . The noise filtering of a pixel (see Fig. 2) depends on the four
neighbouring pixels in the current frame and the corresponding pixels in the past two filtered
frames. In order to support MB based noise filtering, we handle the boundary pixels in a MB
differently (see Fig. 3). This introduces high frequency artifacts at MB boundaries. These high
frequency artifacts are removed by the encoder's quantizer (a low-pass filter) thereby
providing similar SNR compared to a normal noise filter implementation. Therefore, the
desired objective of MB-based noise filtering is achieved without degradation in SNR.
References:  L. Camiciotti and F. Bruls, Adaptive filtering, WO 01/97509-A1
 Albert van der Werf et al., I.McIC: a single-chip MPEG2 video encoder for
storage, IEEE J. of Solid-State Circuits, vol. 32, no. 11,...