Fast long-term motion estimation for H.264 using multiresolution search

Abstract

In this paper, we propose an adaptive motion search window location algorithm which incorporates multiresolution motion estimation information and the motion vector information predicted froin neighboring encoded blocks. To reduce the computational complexity, we employ the spatial and temporal search range reduction that we proposed "1 [l]. Our proposed fast search is implemented and tested in the emerging H.264 encoder reference software (JMS.Oc. 121). 'Iha proposed motion estimation algoritlun is implemented in the low-complexity mode of motion estimation and mode decision ([3]). We compare the proposed algorithm with the full-search method which checks the distortion values of all the candidates in the given search area. With respect to the fullsearch method. our proposed algorithm has two advantages. First, due to the spatial and temporal search range reduction, our algorithm is faster. Second due to the multiresolution search, our proposed algorithm can use larger motion search range with small additional complexity. The proposed algorithm can significantly reduce the computational complexity of motion estimation with slight increase in bitrate as compared to a full-semh that covers the same motion range over wluch OUT proposed fast search operates. The proposed algorithm provides particularly significant gains for video sequences that contain large motions.

Efficient mode selection for H.264 complexity reduction in a Bayesian framework

“…Similarly, low-complexity motion estimation algorithms for H.264 are described in Refs. [9][10][11][12]. As an example, the authors of Ref.…”

“…As an example, the authors of Ref. [9] propose a fast long-term motion estimation algorithm which uses multi-resolution search and motion vector prediction to adaptively locate the motion search window. Computational savings are achieved by predictive motion window location and search range reduction.…”

Low-complexity skip prediction for H.264 through Lagrangian cost estimation

“…The draft H.264/AVC Joint Model document discusses a number of mechanisms to reduce the complexity of an H.264/AVC encoder [4]. Low-complexity motion estimation algorithms for H.264 are described in [5] and [6], while methods to speed up the mode-decision process of an H.264 encoder are proposed in [7]- [14]. An "early skip mode decision" algorithm for H.264 has been proposed in [15].…”

“…This is typically a complex problem due to the inter-dependencies between MBs. The complexity can be reduced from exponential to linear in the number of MBs by assuming MB independence [17], in general leading to a suboptimal solution (4) where now (5) The mode-selection algorithm incorporated into the JM encoder [18] assumes independence between MBs and estimates a suitable value of for every MB as a function of the current quantizer parameter (QP). The MB mode is selected by first encoding the MB in every mode and then selecting the mode that has the minimum R(D) cost given the estimated Lagrange multiplier .…”

Optimized techniques for video compression and enhancement

“…The motion vector is first obtained at the coarse level using some fast search and then refined gradually in subsequent levels to obtain more accurate motion displacement. One of the particular examples for this category is the multi-resolution motion estimation methods [46][47][48]. In these methods, the motion estimation process is performed at different resolution representations of the video frame using different search strategies.…”