Fast in-band motion estimation with variable size block matching

Abstract

In this paper we propose a Fast motion estimation technique that works in the wavelet domain. The computational cost of the algorithm turns out to be proportional to the linear size of the search window instead of its area. We complete our proposal with a variable-size block-matching scheme in the wavelet domain. We integrated the motion estimation algorithm in a fully-scalable wavelet in-band prediction coder inspired by the IB-MCTF (In-Band Motion Compensation Temporal Filtering) proposed in [I].Comparative tests prove that our coder provides the same quality level than IB-MCTF at a very reduced computational cost. Moreover, although our method turns out to match the performance of MCTF-EZBC [2] in terms of PSNR, it clearly outperforms it in terms of perceptual quality. as it is completely free from blocking artifacts.

Fast multiresolution motion estimation algorithms for wavelet-based scalable video coding

“…To demonstrate the performance of the proposed algorithms, we compare the proposed WMEC-PDS algorithm with Spiral-PDS [6] and modified CPME-PDS [17], and the proposed ADCS algorithm with modified FMRME [12] and FIBME [11]. Table 3 gives the expansions of the acronyms of different fast algorithms and also briefly describes their main characteristics.…”

“…The proposed PDS algorithm based on the wavelet matching error characteristic FME FMRME [12] The fast multiresolution motion estimation algorithm which exploits the correlation among the subbands' MVs FIBME [11] The fast in-band motion estimation algorithm using two separable 1D directional displacement refinement ADCS…”

“…However, the coarse-scaled motion vectors for some blocks may not be accurate enough and could cause some errors which propagate along the hierarchical structure. For example, the FIBME [11] method uses a coarse full search in subband LL to determine an initial search point. However, this algorithm tends to give a bad estimate of the motion displacement in case of large displacement.…”

“…Instead of full search in LL subband, a coarse full search in critically sampled LL subband is used in the FIBME [11] method to determine an initial search point. According to the analysis of the computational complexity in the FIBME, the speed-up ratio of the FIBME approximates to 6W with respect to the exhaustive FSA.…”

“…However, these fast motion algorithms do not overcome the shiftvariant property of the wavelet transform because all these algorithms use the critically sampled subbands of the wavelet reference frame for motion estimation and compensation. Maestroni et al [11] proposed a fast in-band motion estimation algorithm (FIBME) that works in shift-invariant wavelet domain. This approach uses a coarse full search in the LL subband to determine an initial motion vector candidate.…”