**6. Experimental results**

In our block based method, we have fixed the Diamond Search as a block searching strategy and the MSE as a block matching criterion since it gives better compression performance while not sacrificing image quality. We have also fixed the size of the window to 7 and the size of the block to 2 since we work in the approximation in the third level of the DWT. Furthermore, we have integrated all the techniques mentioned previously with a quarter of pixel precision and a refinement technique by moving to lower DWT level to re-estimate the poorly predicted blocks.

Our method has proved its performance and robustness for several video benchmarks used to test the ME/MC methods such as the "Tennis", "Foreman", "Susie", "Claire" sequences and even the "Football" sequence which contains large movements.


Table 4. PSNR of the reconstructed image

The reached results showed large performance in terms of quality of reconstructed frame as shown in Table.4 and also in terms of compression ratio. All this, is due to the accuracy of the estimation and the corrections made for the motion vectors.

Our experiments verify the superiority of the proposed ME system, not only versus several other well-known ME systems in the frequency and the multiresolution domains, but also versus the ME systems in the spatial domain. Moreover, it is faster than other methods and the compression ratio is highly increased because it works on the approximation level of the DWT, which is 8 times smaller than the original image.

Furthermore, it is clear from the Figure.11 that there is a big difference between the visual qualities of the reconstructed frames using these different ME/MC systems. We can

All these techniques have united to improve our methods which make it fast, efficient and accurate. In addition, we can even exploit the human visual system and remove the small variations not recognized by the human eye between the two frames. The motion vectors and the prediction error can be encoded after transformed by DWT using the Embedded Zerotree wavelet algorithm (EZW) developed by Shapiro (Shapiro, 1993) or by the Set Partitioning in Hierarchical Trees Algorithm (SPIHT) developed by Said and Pearlman (Said, 1996) which are algorithms that exploit the wavelet structure for an

In our block based method, we have fixed the Diamond Search as a block searching strategy and the MSE as a block matching criterion since it gives better compression performance while not sacrificing image quality. We have also fixed the size of the window to 7 and the size of the block to 2 since we work in the approximation in the third level of the DWT. Furthermore, we have integrated all the techniques mentioned previously with a quarter of pixel precision and a refinement technique by moving to lower DWT level to re-estimate the

Our method has proved its performance and robustness for several video benchmarks used to test the ME/MC methods such as the "Tennis", "Foreman", "Susie", "Claire" sequences

**Methods Tennis Foreman Susie Claire** 

**Spatial domain** 34.3983 33.5550 36.6450 37.7992

**DCT domain** 28.2568 31.3646 31.2833 33.0233

**Conventional DWT** 31.7586 31.2889 33.1613 32.5908

**Proposed method 35.6263 34.6025 38.3417 38.5418** 

The reached results showed large performance in terms of quality of reconstructed frame as shown in Table.4 and also in terms of compression ratio. All this, is due to the accuracy of

Our experiments verify the superiority of the proposed ME system, not only versus several other well-known ME systems in the frequency and the multiresolution domains, but also versus the ME systems in the spatial domain. Moreover, it is faster than other methods and the compression ratio is highly increased because it works on the approximation level of the

Furthermore, it is clear from the Figure.11 that there is a big difference between the visual qualities of the reconstructed frames using these different ME/MC systems. We can

and even the "Football" sequence which contains large movements.

the estimation and the corrections made for the motion vectors.

DWT, which is 8 times smaller than the original image.

efficient coding.

**6. Experimental results** 

poorly predicted blocks.

 **Sequence** 

Table 4. PSNR of the reconstructed image

observe that when the motion estimation is applied on the DCT domain, block effects appeared. On the other hand, using the classical DWT domain, there are also blocks effects, despite its superiority to the DCT domain. Our method gives a better visual quality that resembles to the quality of the reconstructed frame by the spatial domain based ME/MC system.

Fig. 11. The ME/MC results on the 129th frame of the "foreman" sequence. (a) The original image. The estimated frame: (b) ME/MC in the DCT domain, (c) ME/MC in the DWT domain, (d) with our method.

The efficiency of our motion estimation method is well confirmed by the results, in the visual qualities of the reconstructed frames, reached by applying the ME/MC on the Tennis sequence conducted in several domains. The results mentored in Figure.12 consolidate the fact that our motion estimation method outperforms other motion estimations methods conducted in different domains.

Wavelet Transform Based Motion Estimation and Compensation for Video Coding 39

The authors would like to acknowledge the financial support of this work by grants from the General Direction of Scientific Research (DGRST), Tunisia, under the ARUB program.

Arvanitidou, M.G. et al. (2009). Global motion estimation using variable block sizes and its

Barjatya, A. (2004). Block Matching Algorithms For Motion Estimation, DIP 6620 Spring

http://read.pudn.com/downloads153/sourcecode/graph/texture\_mapping/6759

Ben Aoun, N., El'Arbi, M. & Ben Amar, C (2010). Multiresolution motion estimation and

Cheng, P.Y, Li, J. & Jay Kuo, C.-C. (1995). Multiscale video compression using wavelet

Conklin, G.J., Hemami, S.S. (1997). Multi-resolution motion estimation, *International* 

A. Criminisi, A., Cross, G., Blake, A. & Kolmogorov, V. (2006). Bilayer Segmentation of Live

Gharavi, H. & Mills, M. (1990). Block Matching Motion Estimation Algorithms: New Results.

Kutil, R. (2003). Evaluation of wavelet domain block motion compensation (WBMC),

Lundmark, A., Li, H. & Forchheimer, R. (2000). Motion vector certainty reduces bit rate in

Meyer, F.G, Averbuch A. & Coifman R.R. (1997). Motion compensation of wavelet

Nosratinia, A. & Orchard, M.T. (1995). Multi-resolution backward video coding, *International* 

Vol.1 , ISBN 0-7695-2597-0, New York, NY, USA, June 17-22, 2006.

18/BlockMatchingAlgorithmsForMotionEstimation.PDF

8186-7919-0, Munich, Bavaria, Germany, April 21-24, 1997

2004 Final Project Paper, Available from:

1990), pp. 649-651, ISSN 0098-4094

Malo, France, April 23-25, 2003

DC, USA, October 26-29, 1997

D.C., USA, October 23-26, 1995.

Perth, AUSTRALIE, June 20-23, 2000

October 24-28, 2010

1995.

application to object segmentation, 10th *Workshop on Image Analysis for Multimedia Interactive Services*, pp. 173-176, ISBN 978-1-4244-3609-5, London, UK, May 6-8, 2009

compensation for video coding, Proceedings of the 10th International Conference on Signal Processing, Part II, pp.1121-1124, ISBN 978-1-4244-5897-4, Beijing, China,

transform and motion compensation, *International Conference on Image Processing*, pp. 606–609, Vol.1, ISBN 0-8186-7310-9, Washington D.C., USA, October 23-26,

*Conference on Acoustics, Speech, and Signal Processing*, pp. 2873–2876, Vol.4, ISBN 0-

Video, *International Conference on Computer Vision and Pattern Recognition*, pp. 53-60,

*IEEE Transactions on Circuits and Systems for Video Technology*, Vol.37, No.5, (May

*Proceedings of the International Picture Coding Symposium (PCS'03)*, pp. 513-518, Saint

backward motion estimation video coding, *SPIE Proceedings of Visual Communications and Image Processing*, Vol.4067, pp. 95–104, ISBN 0-8194-3703-4,

coefficients for very low bit rate video coding, *Proceedings of IEEE International Conference on Image Processing*, Vol.3, pp. 638-641, ISBN 0-8186-8183-7, Washington,

*Conference on Image Processing*, pp. 563–566, Vol.2, ISBN 0-8186-7310-9, Washington

**8. Acknowledgment** 

**9. References** 

Fig. 12. The ME/MC results on the 17th frame of the "Tennis" sequence. (a) The original image. The estimated image: (b) ME/MC in the DCT domain, (c) ME/MC in the DWT domain, (d) with our method.

## **7. Conclusion**

Video coding has received an increased interest because of the big growth in the quantity of the video data. That is why a big interest has been made for developing an efficient video coding system and improving the motion estimation part which represents the most important part since it consumes most computation time and most resources used for video coding. Making the motion estimation a fast and efficient process was the goal of many researchers. But, unfortunately, that was not reached in the spatial domain. That's why, new ME systems have been conducted in other domain such as the frequency and the multiresolution domain. That is why many studies have been made to improve and simplify the ME methods. In this chapter, we have studied the wavelet as a domain for ME and we have proposed a multiresolution motion estimation and compensation method based on block matching applying in the wavelet coefficients. Because of some problems presented in this chapter, we have integrated some improvements techniques to ameliorate our ME system. As a future works, we will reinforce our method with others techniques such as the spatial segmentation which makes the estimation more accurate by trying to identify real objects in the predicted moving zones.
