Düşük işlem yüküne sahip hareket kestirimi için tümlev imge temelli ikilileştirme

Öz

Günümüzde yüksek çözünürlüklü televizyonlar, kameralar, akıllı telefonların kullanımı ile birlikte yüksek çözünürlüklü video uygulamalarına talep duyulmaktadır. Bu cihazlardaki güç tüketimi ve sınırlı hafıza gibi kısıtlardan dolayı da düşük işlem yüküne sahip video kodlama yöntemlerine ihtiyaç artmaktadır. Video kodlama standartlarında halen en fazla işlem yükü hareket kestirimi kısmındadır.  Bu çalışmada düşük işlem yüküne sahip, düşük bit derinliği gösterimi temelli bir hareket kestirimi yöntemi önerilmektedir. Bu yaklaşımda video çerçeveleri tümlev imge kullanılarak etkin bir şekilde ikilileştirilmekte ve video çerçevelerinin iki bit ile gösterimi elde edilmektedir. Elde edilen ikili çerçeveler üzerinden geleneksel mutlak farklar toplamı (SAD) yerine donanıma daha uygun olan dışaran veya (EX-OR) operasyonu kullanılarak uyumlama işlemi yapılmaktadır. Hareket kestiriminde ikilileştirme işlemi gerçekleştirirken tümlev imge kullanılması ilk kez bu çalışmada önerilmektedir. Önerilen yöntem, literatürde mevcut olan 1-bit dönüşüm (1BT) temelli yaklaşımlara kıyasla hareket kestirim doğruluğunu geliştirirken iki‑bit dönüşüm temelli yaklaşımların başarısı ile hemen hemen aynı seviyede olmaktadır. Bunun yanında özellikle ikilileştirme aşamasında bu yöntemlerin işlem yükünü azaltmaktadır.

Anahtar Kelimeler

• Tümlev imge,İkilileştirme,Düşük bit derinliği,Hareket kestirimi

Integral image based binarization for low-complexity motion estimation

Abstract

Today, high resolution video applications are demanded with the use of high resolution televisions, cameras, smart phones. The requirement for low processing load video coding methods increase due to constraints such as power consumption and limited memory in these devices. In video coding standards, most of the processing load still originates from the motion estimation part. In this study, a low bit‑depth representation based motion estimation method that has low computational load is proposed. In this approach, video frames are binarized efficiently by using integral image and the representation of video frames in terms of two bits is performed. Matching operation is carried out on these binary image frames using hardware‑friendly EX-OR operation instead of conventional SAD (Sum of Absolute Difference This study is the first attempt of using the integral image for the binarization process in the motion estimation. While the proposed method improves the motion estimation accuracy compared to the 1BT based approaches available in the literature, it provides similar motion estimation performance with the two‑bit depth based approaches. Additionally, it reduces the processing load of these methods, especially during the binarization phase.

Keywords

• Integral image,Binarization,Low bit-depth,Motion estimation

Kaynakça

1. Medhat A, Shalaby A, Sayed MS, Elsabrouty M. “A highly parallel SAD architecture for motion estimation in HEVC encoder”. IEEE Asia Pacific Conference Circuits Systems (APCCAS), Ishigaki, 17-20 November 2014.
2. Koga T, Linuma K, Hirano A, Lijima Y, Ishiguro T. “Motion compensated interframe coding for video conferencing”. In National Telecommunication Conference, C9.6.1-C9.6.5, 29 November -3 December 1981.
3. Renxiang L, Bing Z, Ming LL. “A new three-step search algorithm for block motion estimation”. IEEE Transactions Circuits and Systems for Video Technology, 4(4), 438-442, 1994.
4. Lai-Man P, Wing-Chung M. “A novel four-step search algorithm for fast block motion estimation”. IEEE Transactions Circuits and Systems for Video Technology, 6(3), 313-317, 1996.
5. Zhu S, Ma KK. “A new diamond search algorithm for fast block-matching motion estimation”. IEEE Transactions. Image Processing., 9(2), 287-290, 2000.
6. Zhu, C, Lin, X, Chau, LP. “Hexagon-based search pattern for fast block motion estimation. IEEE Transactions Circuits and Systems for Video Technology, 12(5), 349-355, 2002.
7. Yu Y, Zhou J, Chen CW. "A novel fast block motion estimation algorithm based on combined subsamplings on pixels and search candidates". Journal Visual Communication Image Representation, 12(1), 96-105, 2001.
8. Wang CN, Yang SW, Liu CM, Chiang T. “A hierarchical decimation lattice based on n-queen with an application for motion estimation”. IEEE Signal Processing Letters, 10(8), 228-231, 2003.

9. Cheung CK, Lai-Man P. “Normalized partial distortion search algorithm for block motion estimation”. IEEE Transactions Circuits and Systems for Video Technology,10(3), 417-422, 2000.
10. Kim I, Kim J, Jeong J, Jeon G. “Low-complexity block-based motion estimation algorithm using adaptive search range adjustment”. Optical Engineering, 51(6), 2012.
11. Li W, Salari E. “Successive elimination algorithm for motion estimation”. IEEE Transactions Image Processing, 4(1), 105-107, 1995.
12. Natarajan B, Bhaskaran V, Konstantinides K. "Low-complexity block-based motion estimation via one-bit transforms". IEEE Transactions Circuits and Systems for Video Technology, 7(4), 702-706, 1997.
13. Ertürk S. "Multiplication-Free one-bit transform for low-complexity block-based motion estimation". IEEE Signal Processing Letters, 14(2), 109-112, 2007.
14. Ertürk A, Ertürk S. "Two bit transform for binary block motion estimation". IEEE Transactions Circuits and Systems for Video Technology, 15(7), 938-946, 2005.
15. Urhan O, Ertürk S. “Constrained one-bit transform for low-complexity block motion estimation”. IEEE Transactions Circuits and Systems for Video Technology, 17(4), 478-482, 2007.
16. Choi C, Jeong J. ‘Enhanced two-bit transform based motion estimation via Extension of matching criterion”. IEEE Transactions Consumer Electronics, 56(3), 1883-1889, 2010.
17. Güllü MK. "Weighted constrained one-bit transform based fast block motion estimation". IEEE Transactions Consumer Electronics, 57, 751-755, 2011.
18. Song CM, Guo Y, Wang XH, Liu D. "Fuzzy quantization based bit transform for low bit-resolution motion estimation". Signal Processing: Image Communication, 28(10), 1435-1447, 2013.
19. Sankur B, Sezgin M. "A survey over ımage thresholding techniques and quantitative performance evaluation". Journal of Electronic Imaging, 13(1), 146-165, 2004.
20. Sauvo la J, Pietikainen M. “Adaptive document image binarization”. Pattern Recognition, 33(2), 225-236, 2000.
21. Sauvola J, Seppanen T, Haapakoski S, Pietikainen M. "Adaptive document binarization". 4th International Conference. on Document Analysis and Recognition, Ulm Germany, 18-20 August 1997.
22. Romen ST, Roy S, Imocha SO, Sinam T, Manglem SK. "A new local adaptive thresholding technique in binarization". International Journal of Computer Science, 6( 2), 271-277, 2011.
23. Crow F. "Summed-area tables for texture mapping". In Proceedings of SIGGRAPH, July 1984.
24. Viola P, Jones MJ. “Robust real-time face detection”. International Journal of Computer Vision, 57(2), 137-154, 2004.