PATCH-ASOSLANGAN MRF–BP USULI ASOSIDA YUZNI FRONTALIZATSIYA QILISH: PRIORITY BELIEF PROPAGATION VA FOURIER DOMENIDA TEZLASHTIRILGAN ALGORITM

Authors

  • F. Maxmudxo‘jayev Author
  • Boqiyev Farrux Farrux Author

Keywords:

yuzni frontalizatsiya qilish, Markov tasodifiy maydonlari, Belief Propagation, patch-asoslangan moslashtirish, Lucas–Kanade algoritmi, Gabor filtrlari, Fourier transformatsiyasi, kompyuterli ko‘rish, biometrik identifikatsiya.

Abstract

Mazkur maqolada real shart-sharoitlardagi yuzni aniqlash tizimlarining samaradorligini sezilarli darajada cheklab qo‘yadigan pozitsiya muammosini hal etishga qaratilgan yangi frontalizatsiya usuli taklif etilgan. Patch-asoslangan Markov tasodifiy maydonlari (MRF) va Belief Propagation (BP) optimizatsiyasiga asoslangan klassik yondashuv ustuvor xabarlarni rejalashtirish (Priority BP), dinamik yorliqlarni kesish (label pruning) va Fourier domenida tezlashtirilgan= Lucas–Kanade (LK) algoritmi bilan boyitilgan. FERET, CMU-PIE va Multi-PIE xalqaro etalon ma’lumotlar bazalarida o‘tkazilgan eksperimental sinov natijalari taklif etilayotgan yondashuvning klassik MRF–BP usulidan o‘rtacha 3.9% ga, hatto generativ adversarial tarmoq TP-GAN dan 1.2% ga yuqori Rank-1 aniqlash darajasi bilan ishlay olishini ko‘rsatdi. Tizim oddiy ofis kompyuterida 1.5 sekundda frontalizatsiya operatsiyasini bajara oladi va 500 MB RAM dan kam resurs talab qiladi.

References

1. Schroff F., Kalenichenko D., Philbin J. FaceNet: A unified embedding for face

recognition and clustering // Proc. IEEE CVPR. – 2015. – P. 815–823.

2. Parkhi O. M., Vedaldi A., Zisserman A. Deep face recognition // Proc. BMVC. –

2015. – P. 1–12.

3. Deng J., Guo J., Xue N., Zafeiriou S. ArcFace: Additive angular margin loss for

deep face recognition // Proc. IEEE CVPR. – 2019. – P. 4690–4699.

4. Blanz V., Vetter T. A morphable model for the synthesis of 3D faces // Proc.

SIGGRAPH. – 1999. – P. 187–194.

5. Kanade T., Yamada A. Multi-subregion based probabilistic approach toward

pose-invariant face recognition // Proc. CIRA. – 2003. – P. 954–959.

6. Arashloo S. R., Kittler J. Energy normalization for pose-invariant face

recognition based on MRF model image matching // IEEE Transactions on

PAMI. – 2011. – Vol. 33, No. 6. – P. 1274–1280.

7. Tran L., Yin X., Liu X. Disentangled representation learning GAN for poseinvariant face recognition // Proc. IEEE CVPR. – 2017. – P. 1415–1424.

8. Huang R., Zhang S., Li T., He R. Beyond face rotation: Global and local

perception GAN for photorealistic and identity preserving frontal view

synthesis // Proc. ICCV. – 2017. – P. 2439–2448.

9. Komodakis N., Tziritas G. Image completion using efficient belief propagation

// IEEE Transactions on Image Processing. – 2007. – Vol. 16, No. 11. – P.

2649–2661.

10. Ashraf A. B., Lucey S., Chen T. Fast image alignment in the Fourier domain //

Proc. IEEE CVPR. – 2010. – P. 2480–2487.

11. Lucas B. D., Kanade T. An iterative image registration technique with an

application to stereo vision // Proc. IJCAI. – 1981. – P. 674–679.

12. Phillips P. J., Moon H., Rizvi S. A., Rauss P. J. The FERET evaluation

methodology for face-recognition algorithms // IEEE Transactions on PAMI. –

2000. – Vol. 22, No. 10. – P. 1090–1104.

13. Sim T., Baker S., Bsat M. The CMU pose, illumination, and expression

database // IEEE Transactions on PAMI. – 2003. – Vol. 25, No. 12. – P. 1615–

1618.

14. Gross R., Matthews I., Cohn J., Kanade T., Baker S. Multi-PIE // Proc. IEEE

FG. – 2008. – P. 807–813.

15. Zhang W., Shan S., Gao W., Chen X., Zhang H. Local Gabor binary pattern

histogram sequence (LGBPHS) // Proc. ICCV. – 2005. – P. 786–791.

16. Chai X., Shan S., Chen X., Gao W. Locally linear regression for pose-invariant

face recognition // IEEE Transactions on Image Processing. – 2007. – Vol. 16,

No. 7. – P. 1716–1725.

17. Asthana A., Marks T. K., Jones M. J., Tieu K. H. Fully automatic poseinvariant face recognition via 3D pose normalization // Proc. ICCV. – 2011. –

P. 937–944.

18. Liu X., Cao Y., Li L. et al. DiffFace: Diffusion-based face swapping with

identity preservation // arXiv preprint arXiv:2304.13715. – 2023.

Published

2026-05-08