Tài liệu Nội suy và tăng cường chất lượng video sử dụng máy học

ĈҤ,+Ӑ&48Ӕ&*,$73+CM 75ѬӠ1*ĈҤ,+Ӑ&%È&+.+2$ 1*8<ӈ11*Ӑ&;8Æ1+8< 1Ӝ,68<9¬7Ă1*&ѬӠ1*&+Ҩ7/ѬӦ1*9,'(2 6Ӱ'Ө1*0È<+Ӑ& Chuyên ngành: .ӻ7KXұW9LӉQ7K{QJ 0mVӕ 60.52.02.08 /8Ұ19Ă17+Ҥ&6Ƭ 73+Ӗ&+Ë0,1+WKiQJ08 QăP Công trình ÿѭӧc hoàn thành tҥi: 7UѭӡQJĈҥi hӑc Bách Khoa ± Ĉ+4*-HCM Cán bӝ Kѭӟng dүn khoa hӑc: PGS.TS. Hà Hoàng Kha Cán bӝ chҩm nhұn xét 1 : GS. TS. Lê TiӃQ7Kѭӡng Cán bӝ chҩm nhұn xét 2 : TS. NguyӉQĈuQK/RQJ Luұn YăQWKҥFVƭÿѭӧc bҧo vӋ tҥL7UѭӡQJĈҥi hӑF%iFK.KRDĈ+4*73HCM QJj\WKiQJQăP (trӵc tuyӃn). Thành phҫn HӝLÿӗQJÿiQKJLiOXұQYăQWKҥFVƭJӗm: 1. Chӫ tӏFK3*676Ĉӛ Hӗng Tuҩn 7KѭNê76+XǤQK3K~0LQK&ѭӡng 3. Phҧn biӋn 1: GS. TS. Lê TiӃQ7Kѭӡng 4. Phҧn biӋn 2: TS. NguyӉQĈuQK/RQJ 5. Ӫy viên: PGS. TS. Võ NguyӉn Quӕc Bҧo Xác nhұn cӫa Chӫ tӏch HӝLÿӗQJÿiQKJLi/9Yj7Uѭӣng Khoa quҧn lý chuyên ngành sau khi luұQYăQÿã ÿѭӧc sӱa chӳa (nӃu có). CHӪ TӎCH HӜ,ĈӖNG 75ѬӢNG KHOA Ĉ,ӊN - Ĉ,ӊN TӰ ĈҤ,+Ӑ&48Ӕ&*,$73+&0 75ѬӠ1*ĈҤ,+Ӑ&%È&+.+2$ &Ӝ1*+Ñ$;­+Ӝ,&+Ӫ1*+Ƭ$9,ӊ71$0 ĈӝFOұS- 7ӵGR- +ҥQKSK~F 1+,ӊ09Ө/8Ұ19Ă17+Ҥ&6Ƭ +ӑWrQKӑFYLrQ1JX\ӉQ1JӑF;XkQ+X\ ............................... MSHV: 1770336 1Jj\WKiQJQăPVLQK04/11/1993 ........................................... 1ѫLVLQK%ҥF/LrX &KX\rQQJjQK.ӻWKXұWYLӉQWK{QJ ......................................... 0mVӕ : 60 52 02 08 I. 7Ç1Ĉӄ7¬, 7LӃQJ9LӋW1ӝLVX\YjWăQJFѭӡQJFKҩWOѭӧQJYLGHRVӱGөQJPi\KӑF 7LӃQJ $QK Interpolation and Enhancement of Video Quality using Machine Learning II. 1+,ӊ09Ө9¬1Ӝ,'81* - - - 1JKLrQFӭu FiFJLҧLSKiSQӝLVX\YjWăQJFѭӡQJFKҩWOѭӧQJYLGHRVӱGөQJ Pi\KӑF7URQJÿyWұSWUXQJYjRQӝLGXQJFKtQK7KrPPjXFKRYLGHRWUҳQJ ÿHQWăQJÿӝSKkQJLҧLYjWăQJIUDPHUDWH 7KDPNKҧRVRViQKQKұQ[pWYӟLFiFF{QJWUuQKQJKLrQFӭXWUѭӟFÿy ĈѭDUD PӝWJLҧLSKiSKӋWKӕQJÿӇFҧLWLӃQKRһFWӕLѭXFKREjLWRiQGӵDWUrQQӅQWҧQJ Pi\KӑF. 7KӵFQJKLӋPFiFJLҧLSKiSKLӋQWKӵFKyDFKѭѫQJWUuQKtrên máy tính Google Colab. ĈѭDUDEҧQJÿiQKJLiNӃWTXҧ III. NGÀY *,$21+,ӊ09Ө : *KLWKHRWURQJ4ĈJLDRÿӅWjL 22/02/2021 IV. 1*¬<+2¬17+¬1+1+,ӊ09Ө*KLWKHRWURQJ4ĈJLDRÿӅWjL 13/06/2021 V. &È1%Ӝ+ѬӞ1*'Ү1*KLU}KӑFKjPKӑFYӏKӑWrQ PGS. TS. HÀ HOÀNG KHA 7S+&0QJj\WKiQJQăP21 CÁN BӜ +ѬӞNG DҮN (Hӑ tên và chӳ ký) CHӪ NHIӊM BӜ 0Ð1Ĉ¬27ҤO (Hӑ tên và chӳ ký) 75ѬӢNG KHOA Ĉ,ӊN - Ĉ,ӊN TӰ (Hӑ tên và chӳ ký) /ͤ,&̪0ˮ1 Ĉ̯XWLrQW{L[LQÿ˱ͫc g͵i lͥi c̫P˯QFKkQWKjQKÿ͇n th̯\K˱ͣng d̳n lu̵QYăQ cͯa mình PGS.TS. Hà Hoàng Kha. Trong quá trình th͹c hi͏n lu̵QYăQWK̯\OjQJ˱ͥi ÿmQKL͏t tình h͟ trͫ, ch͑ d̳n giúp tôi cͯng c͙ ki͇n thͱFÿ͛ng thͥi ch͑ ra nhͷng v̭n ÿ͉ c͙t lõi giúSW{LFyÿ͓QKK˱ͣQJÿ~QJÿ̷Qÿ͋ hoàn thành lu̵QYăQ Ti͇Sÿ͇QW{L[LQÿ˱ͫc g͵i lͥi c̫P˯Qÿ͇n các th̯\F{ÿmYjÿDQJG̩y t̩LWU˱ͥng Ĉ̩i H͕c Bách Khoa Tp.HCM Yjÿ̿c bi͏t là các th̯y cô ͧ B͡ Môn Vi͍n Thông ÿm giúp tôi xây d͹QJÿ˱ͫc ki͇n thͱc n͉n t̫QJOjF˯ sͧ ÿ͋ tôi th͹c hi͏Qÿ˱ͫc lu̵QYăQ này. Cu͙i cùng tôi xin g͵i lͥi c̫P˯Qÿ͇n JLDÿuQKE̩QEqÿ͛ng nghi͏p ÿmK͇t sͱc JL~SÿͩTXDQWkPÿ͡QJYLrQÿ͋ W{LFyÿL͉u ki͏n thu̵n lͫLÿ͋ th͹c hi͏n lu̵QYăQQj\ Xin chân thành c̫P˯Q Tp. H͛ Chí Minh, ngày tháng QăP21 NGUY͌N NG͔C XUÂN HUY 1ӝLVX\YjWăQJFѭӡQJFKҩWOѭӧQJYLGHRVӱGөQJPi\KӑF GVHD: PGS.TS. Hà Hoàng Kha 7Ï07Ҳ7/8Ұ19Ă1 1Jj\QD\YLGHRÿDQJGҫQWUӣWKjQKPӝWNrQKFXQJFҩSWK{QJWLQWUӵFTXDQWKӕQJ WUӏWUrQFiFPҥQJ[mKӝLKѫQFҧWK{QJWLQGҥQJYăQEҧQKRһFKuQKҧQK9LӋFF{QJ QJKӋQJj\FjQJQkQJFҩSFKҩWOѭӧQJFiFEӝKLӇQWKӏÿzLKӓLFyPӝWJLҧLSKiSGjQK FKRFiFQJXӗQWK{QJWLQҧQKYjYLGHROkXÿӡLFyFKҩWOѭӧQJWKҩSVRYӟLFiFWLrXFKXҭQ KLӋQWҥLFyWKӇÿiSӭQJÿѭӧFYӟLVӵQkQJFDRFKҩWOѭӧQJQj\'RYұ\FiFNӻWKXұW WăQJFѭӡQJFKҩWOѭӧQJKuQKҧQKYjYLGHRQJj\FjQJÿѭӧFFK~WUӑQJ1yLPӝWFiFK ÿѫQJLҧQFiFNӻWKXұWWăQJFѭӡQJOjPӝWTXiWUuQKÿӇFҧLWKLӋQFKҩWOѭӧQJFӫDYLGHR ӣPӝWVӕNKtDFҥQKQjRQKҩWÿӏQKÿyFyWKӇOjFKX\ӇQQӝLGXQJWUҳQJÿHn thành có PjXVҳFWăQJÿӝ SKkQJLҧLFҧLWKLӋQFKҩWOѭӧQJSKѫLViQJFkQEҵQJYLGHRJLҧP QKLӉXWăQJWӕFÿӝIUDPHKRһFWKrPFiFKLӋXӭQJĈk\OjSKҫQFҫQWKLӃWQKҩWWURQJ EҩWNǤPӝWKӋWKӕQJ[ӱOêWăQJFѭӡQJQjR9uYұ\PөFWLrXFӫDÿӅWjLOjWұSWUXng JLҧLTX\ӃWEjLWRiQ³1ӝLVX\YjWăQJFѭӡQJFKҩWOѭӧQJYLGHR´GӵDWUrQQӅQWҧQJKӑF Pi\YjWұSWUXQJYjRNKtDFҥQKWăQJWӕFIUDPHUDWHWăQJFѭӡQJÿӝSKkQJLҧLYjQӝL VX\PjXVҳF 9ӅFҩXWU~FQJKLrQFӭXOXұQYăQVӁJLӟLWKLӋXYӅ[XKѭӟQJJLDWăQJQKXFҫXVӱ GөQJFiFEjLWRiQWăQJFѭӡQJFKҩWOѭӧQJFKRYLGHRYj[XKѭӟQJFiFQJKLrQFӭXYӅ YҩQÿӅQj\WURQJWKӡLJLDQJҫQÿk\FK~WUӑQJYjRFiFSKѭѫQJSKiSVӱGөQJKӑFPi\ /XұQYăQÿѭDUDFiFQKLӋPYөYjSKҥPYLQJKLrQFӭXÿӗQJWKӡLÿiQKJLiѭXQKѭӧF ÿLӇPFiFSKѭѫQJSKiSÿmFy7ӯÿyFKӑQUDFiFWLrXFKtÿӇOӵDFKӑQSKѭѫQJSKiS WăQJFѭӡQJFKRYLGHRYjFiFKѭӟQJ QJKLrQFӭXSKKӧS&iFSKҫQWLӃSWKHROҫQOѭӧW QrXUDFѫVӣOêWKX\ӃWOLrQTXDQEDRJӗPFiFP{KuQKJLҧLWKXұWÿѭӧFVӱGөQJFNJQJ QKѭFKҥ\WKӱWUrQ*RRJOH&RODE&XӕLFQJOXұQYăQWUuQKEj\NӃWTXҧQJKLrQFӭX YjWKӱQJKLӋPFiFGӵNLӃQÿyQJJySFӫD ÿӅWjLNӃWOXұQYjFiFKѭӟQJSKiWWULӇQ WURQJWѭѫQJODL 7yPWҳWOXұQYăQ i +91JX\ӉQ1JӑF;XkQ+X\ 1ӝLVX\YjWăQJFѭӡQJFKҩWOѭӧQJYLGHRVӱGөQJPi\KӑF GVHD: PGS.TS. Hà Hoàng Kha ABSTRACT Today, video is gradually becoming a dominant visual channel to information storage or transmission, more than texts or images. As technology increasingly improves the quality of display units, a solution for long-standing, low-quality photo and video sources compared to current standards is required so that it can be met with today quality improvement. Therefore, techniques to enhance image and video quality are significantly focused. Simply put, enhancement techniques are a process to improve the quality of a video in certain aspects, that can be turning black and white content into color, increasing resolution, improving exposure quality, balancing video, reducing noise, increasing frame rate or adding effects, etc. This is the most essential part of any enhancement processing systems. Therefore, the goal of the study is to focus on solving the problem of "Interpolation and enhancement of YLGHR TXDOLW\ XVLQJ PDFKLQH OHDUQLQJ´ 6SHFLILFDlly, on 3 aspects of framerate acceleration, resolution enhancement and colorization. Regarding this research structure, this thesis will introduce the trend of increasing demand for using quality enhancement problems for video and the trend of researches on this issue in recent times, focusing on methods of using machine learning. This thesis sets out the tasks and research scope, and evaluates the advantages and disadvantages of the available methods. From there, select the criteria to choose the suitable enhancement methods for the video and the appropriate research directions. The following sections, in turn, outline the relevant theoretical background, including models and algorithms used as well as test runs on Google Colab. Finally, the thesis presents the results of the our work on sample videos, the expected contributions of this research, conclusions and future development directions. Abstract ii +91JX\ӉQ1JӑF;XkQ+X\ 1ӝLVX\YjWăQJFѭӡQJFKҩWOѭӧQJYLGHRVӱGөQJPi\KӑF GVHD: PGS.TS. Hà Hoàng Kha /Ӡ,&$0Ĉ2$1 7{L[LQFDPÿRDQÿk\OjF{QJWUuQKQJKLrQFӭXFӫDULrQJW{LGR3*676+j +RjQJ.KDKѭӟng dүn. Các lý thuyӃt, thông sӕ ÿѭӧc trình bày trong luұQYăQÿѭӧc trích dүn tӯ các tài liӋu khoa hӑFÿѭӧc công bӕ bӣi các tә chӭc uy tín. Các kӃt quҧ tính toán và thӵc nghiӋm ÿѭӧc trình bày trong luұQYăn là kӃt quҧ trung thӵc và do tôi thӵc hiӋn. Tp. Hӗ Chí Minh, ngày tháng QăP21 NguyӉn Ngӑc Xuân Huy /ӡLFDPÿRDQ iii +91JX\ӉQ1JӑF;XkQ+X\ 1ӝLVX\YjWăQJFѭӡQJFKҩWOѭӧQJYLGHRVӱGөQJPi\KӑF GVHD: PGS.TS. Hà Hoàng Kha 0Ө&/Ө& DANH 0Ө& CÁC 7Ӯ 9,ӂ7 7Ҳ7 VI DANH 0Ө& HÌNH Ҧ1+ VII DANH 0Ө& %Ҧ1* %,ӆ8 IX 1 MӢĈҪ8 1 1.1 /êGRFKӑQÿӅWjL 2 1.2 0өFWLrXQJKLrQFӭX 4 1.3 ĈӕLWѭӧQJYjSKҥPYLQJKLrQFӭX 4 1.4 3KѭѫQJSKiSQJKLrQFӭX 4 1.5 &ҩXWU~FOXұQYăQ 5 2 G,Ӟ,7+,ӊ87Ә1*48$1 6 2.1 *LҧLWKtFKPӝWVӕNKiLQLӋPYLGHRYjFiFYҩQÿӅÿһWUD 2.2 &iFQJKLrQFӭXOLrQTXDQWӟLQkQJFDRÿӝSKkQJLҧL 10 2.3 &iFQJKLrQFӭXOLrQTXDQWӟLWKrPPjXFKRҧQKWUҳQJÿHQ 11 2.4 &iFQJKLrQFӭXOLrQTXDQWӟLQkQJFDRWӕFÿӝIUDPHUDWH 13 2.5 .ӃWOXұQ 15 7 3 CѪ6Ӣ/é7+8<ӂ79ӄ;Ӱ/éҦ1+6Ӕ, 9,'(29¬0Ӝ76Ӕ0Ҥ1*1(85$/ 16 3.1 ;ӱOêҧQKVӕYjYLGHR 17 3.2 Convolutional Neural Network (CNN) 21 3.3 Genrative Adversarial Network (GAN) 25 3.4 .ӃWOXұQ 26 4 CÈ&0Ð+Î1+1Ӝ,68<9¬7Ă1*&ѬӠ1*&+Ҩ7/ѬӦ1*9,'(2 27 4.1 7әQJTXDQYӅVѫÿӗNKӕLKӋWKӕQJ 28 4.2 7ăQJWӕFÿӝ framerate 29 4.3 7ăQJFѭӡQJÿӝSKkQJLҧL 33 4.4 1ӝLVX\PjXVҳFFKRYLGHR 38 4.5 .ӃWOXұQ 41 0өFOөF iv +91JX\ӉQ1JӑF;XkQ+X\ 1ӝLVX\YjWăQJFѭӡQJFKҩWOѭӧQJYLGHRVӱGөQJPi\KӑF GVHD: PGS.TS. Hà Hoàng Kha 5 Kӂ748ҦĈҤ7ĈѬӦ&9¬ĈÈ1+*,È 42 5.1 7ULӇQNKDLOXұQYăQYjNӃWTXҧ 43 5.2 1KұQ[pWYjÿiQKJLi 50 6 Kӂ7/8Ұ19¬+ѬӞ1*3+È775,ӆ1 56 6.1 .ӃWOXұQ 57 6.2 +ѭӟQJSKiWWULӇQ 59 T¬,/,ӊ87+$0.+Ҧ2 0өFOөF 61 v +91JX\ӉQ1JӑF;XkQ+X\ 1ӝLVX\YjWăQJFѭӡQJFKҩWOѭӧQJYLGHRVӱGөQJPi\KӑF GVHD: PGS.TS. Hà Hoàng Kha DANH MӨC CÁC TӮ VIӂT TҲT 7ӯYLӃWWҳW 7ӯJӕF 1JKƭDWLӃQJ9LӋW CNN Convolutional Neural Network 0ҥQJQHXUDOWtFKFKұS GAN Generative Adversarial Network 0ҥQJÿӕLQJKӏFK WҥRVLQK FPS Frame per Second .KXQJKuQKPӛLJLk\ CIE Commission Internationale Ӫ\EDQTXӕFWӃ de l'Éclairage YӅFKLӃXViQJ SR Super Resolution ĈӝSKkQJLҧLFDR SRGAN Super Resolution 0ҥQJÿӕLQJKӏFK Generative Adversarial Net WҥRVLQKSKkQJLҧLFDR Motion Estimation ѬӟFOѭӧQJYjE and Motion Compensation FKX\ӇQÿӝQJ MOS Mean Opinion Score ĈLӇPêNLӃQWUXQJEuQK MSE Mean Square Error 6DLVӕWRjQSKѭѫQJ MEMC trung bình SSIM Structure Similarity Index Measure &KӍVӕWѭѫQJÿӗQJ YӅFҩXWU~F PSNR Peak Signal to Noise Ratio 7ӹVӕWtQKLӋXWUrQQKLӉX LR Low Resolution ĈӝSKkQJLҧLWKҩS HR High Resolution ĈӝSKkQJLҧLFDR AuC Raw Accuracy ĈӝFKtQK[iFWK{ LPIPS Learned Perceptual ĈӝWѭѫQJÿӗQJYӅ Image Patch Similarity WKӏJLiF 'DQKPөFFiFWӯYLӃWWҳW vi +91JX\ӉQ1JӑF;XkQ+X\ 1ӝLVX\YjWăQJFѭӡQJFKҩWOѭӧQJYLGHRVӱGөQJPi\KӑF GVHD: PGS.TS. Hà Hoàng Kha DANH MӨC HÌNH ҦNH Hunh 1.1 &KL͇QOͫLSḴPFͯDĈͱFWURQJFKL͇QWUDQKWK͇JLͣLWKͱÿ˱ͫFF̭WJLͷD W̩L6FKORVVNLUFKH(OOLQJHQ%DYDULDWKiQJ 2 Hunh 1.2 7͙Fÿ͡WăQJWU˱ͧQJÿ͡SKkQJL̫LPjQKuQKWURQJWK̵SQLrQV 3 Hunh 1.3 7͙Fÿ͡WăQJWU˱ͧQJGXQJO˱ͫQJWK͓WU˱ͥQJQJjQKVWUHDPLQJYLGHRNKX Y͹FFKkXȱ 7KiL%uQK'˱˯QJWͳ± 2028 3 Hunh 2.1 .tFKWK˱ͣFFͯDP͡WV͙F̫PEL͇QWK{QJGͭQJ 8 Hunh 2.2 +uQKPLQKK͕DW͙Fÿ͡PjQWU̵SFͯDP͡WPi\̫QK 8 Hunh 2.3 1KjTXD\SKLP7KRPDV0HLJKDQWUrQSKLPWU˱ͥQJQăP 9 Hunh 2.4 0͡W̫QKPjXÿ˱ͫFFKͭSWͳEͧL(GZDUG7XUQHU 10 Hunh 2.5 0{W̫EjLWRiQWăQJF˱ͥQJÿ͡SKkQJL̫L 10 Hunh 2.6 ̪QKQ͡LVX\ÿ͡WăQJF˱ͥQJÿ͡SKkQJL̫LFͯDSK˱˯QJSKiSNKiFQKDX VRYͣL̫QKJ͙F 11 Hunh 2.7 9tGͭV͵GͭQJ̫QKWKDPFKL͇XUHIÿ͋Q͡LVX\PjXV̷FFKR̫QK[iP 12 Hunh 2.8 9tGͭV͵GͭQJ³JͫLê´WͳQJ˱ͥLGQJÿ͋Q͡LVX\PjXFKR̫QK[iP 12 Hunh 2.9 0͡WV͙N͇WTX̫Q͡LVX\PjXV̷FV͵GͭQJK͕FPi\ 13 Hunh 2.10 6RViQKFiFSK˱˯QJSKiSQ͡LVX\PjXV̷F 13 Hunh 2.11 9tGͭY͉P͡WP{KuQK[͵OêWăQJW͙Fÿ͡IUDPHG͹DWUrQK͕FPi\FyV͵ GͭQJNKiLQL͏PGHSWKPDSÿ͋WăQJF˱ͥQJFK̭WO˱ͫQJ 14 Hunh 2.12 +L͏QW˱ͫQJQKzHFKX\͋Qÿ͡QJWURQJP͡WV͙SK˱˯QJSKiSQ͡LVX\IUDPH 15 Hunh 3.1 ̪QKWKDQJ[iPFͯD/LQFROQ 17 Hunh 3.2 &iFIUDPHWK͋KL͏QP͡WY̵Qÿ͡QJYLrQF˱ͩLQJ͹D 18 Hunh 3.3 6˯ÿ͛FiFNK{QJJLDQPjXÿmÿ͉F̵S 19 Hunh 3.4 Không gian màu CIE LAB 20 Hunh 3.5 +RDK͛QJYͣLPjXV̷FNKiFQKDXNqPWKHRWK{QJV͙FͯDNK{QJJLDQ màu CIE LAB 21 Hunh 3.6 0{W̫P͡WP̩QJQHXUDOWK{QJWK˱ͥQJ 'DQKPөFKuQKҧQK- EҧQJELӇX vii 22 +91JX\ӉQ1JӑF;XkQ+X\ 1ӝLVX\YjWăQJFѭӡQJFKҩWOѭӧQJYLGHRVӱGөQJPi\KӑF GVHD: PGS.TS. Hà Hoàng Kha Hunh 3.7 0{W̫P͡WP̩QJ&11 22 Hunh 3.8 0D[SRROLQJNtFKWK˱ͣF[ 24 Hunh 3.9 0LQKK͕Dÿ˯QJL̫QP͡WP̩QJQHXUDOGQJÿ͋QK̵QG̩QJFKy 24 Hunh 4.1 /˱Xÿ͛JL̫LWKX̵WQ͡LVX\YjWăQJF˱ͥQJFK̭WO˱ͫQJYLGHR 28 Hunh 4.2 .L͇QWU~Fÿ͉[X̭WWKHRJL̫LWKX̵W'$,1 30 Hunh 4.3 0{W̫P͡WGHSWKPDS̫QKWUiLOj̫QKÿ̯XYjR̫QKSK̫LOjGHSWKPDS FͯD̫QKÿ̯XYjRW˱˯QJͱQJ 30 Hunh 4.4 M{KuQKWUtFK[X̭WWuQKKX͙QJ 32 Hunh 4.5 .L͇Q WU~F FͯD E͡ *HQHUDWRU Yj 'LVFULPLQDWRU FͯD P̩QJ *$1 WURQJ SK˱˯QJSKiS65*$1 35 Hunh 4.6 .͇WTX̫ÿiQKJLi026WUrQE͡GͷOL͏X%SD100 37 Hunh 4.7 7͝QJTXDQY͉NL͇QWU~FFͯDP{KuQK,QVWDQFH-Aware Colorization 38 Hunh 4.8 .L͇QWU~Fÿ͉[X̭WEͧL=KDQJYjF͡QJV͹GQJÿ͋WKrPPjXFKR̫QK[iP 39 Hunh 4.9 .L͇QWU~FFͯDP̩QJPRGXOH)XVLRQ 40 Hunh 5.1 &̭XWU~FP{KuQKOX̵QYăQWK͹FKL͏QWUrQ*RRJOH&RODE 44 Hunh 5.2 6˯ÿ͛PLrXW̫TX\WUuQKV͵GͭQJ'$,1ÿ͋WăQJW͙Fÿ͡IUDPHFKRYLGHR 45 Hunh 5.3 6˯ÿ͛PLrXW̫TX\WUuQK65*$1[U ÿ͋WăQJF˱ͥQJÿ͡SKkQJL̫L cho video 46 Hunh 5.4 6˯ÿ͛PLrXW̫TX\WUuQK,QVWDQFH-$ZDUH&RORUL]DWLRQÿ͋Q͡LVX\PjXV̷F cho video 47 Hunh 5.5 *LiWU͓06(Yj66,0NKLVRViQK̫QKYͣLFKtQKQy 51 Hunh 5.6 *LiWU͓06(Yj66,0NKLVRViQK̫QKFytWQpWW˱˯QJÿ͛QJ 52 Hunh 5.7 *LiWU͓06(Yj66,0FͯDP͡WV͙IUDPHVDXNKLQ͡LVX\ 53 'DQKPөFKuQKҧQK- EҧQJELӇX viii +91JX\ӉQ1JӑF;XkQ+X\ 1ӝLVX\YjWăQJFѭӡQJFKҩWOѭӧQJYLGHRVӱGөQJPi\KӑF GVHD: PGS.TS. Hà Hoàng Kha DANH MӨC BҦNG BIӆU %̫QJ  %̫QJ VR ViQK N͇W TX̫ ÿiQK JLi 3615 Yj 026 FͯD WͳQJ KjP ORVVIXQFWLRQÿ͙LYͣLFiFE͡GͷOL͏X6HWYj6HW 36 %̫QJ %̫QJVRViQKN͇WTX̫ÿiQKJLi3615Yj026FͯDP͡WV͙SK˱˯QJSKiS Q͡LVX\FK̭WO˱ͫQJ̫QK[YͣL E͡GͷOL͏X6HW6HWYj%6' 37 %̫QJ 7K{QJWLQY͉FK̭WO˱ͫQJYLGHRÿ̯XUDYjÿ̯XYjRFͯDK͏WK͙QJNqPYͣL WK{QJWLQY͉JURXQGWUXWKÿ͋ÿiQKJLi 48 %̫QJ 0͡WV͙ IUDPHWU˱ͣFYjVDX[͵OêQ͡LVX\YjWăQJF˱ͥQJWURQJOX̵QYăQ NqPWKHRFiFIUDPHJURXQGWUXWKFͯDQyQ͇XFy 49 %̫QJ %̫QJW͝QJKͫSP͡WV͙WK{QJV͙ÿiQKJLi06(66,0Yj026GjQKFKR FiFYLGHRÿ˱ͫFQ͡LVX\YjWăQJF˱ͥQJFK̭WO˱ͫQJ 53 'DQKPөFKuQKҧQK- EҧQJELӇX ix +91JX\ӉQ1JӑF;XkQ+X\ 1ӝLVX\YjWăQJFѭӡQJFKҩWOѭӧQJYLGHRVӱGөQJPi\KӑF GVHD: PGS.TS. Hà Hoàng Kha &+ѬѪ1* 0ӢĈҪ8 &KѭѫQJ0ӣÿҫXÿӅFұSWӟLFiFYҩQÿӅFѫEҧQFӫDÿӅWjL3KҫQÿҫXVӁÿӅFұS OêGRFKӑQÿӅWjL&өWKӇSKҫQQj\VӁQrXOrQWuQKKuQKSKiWWULӇQFӫDFiFF{QJQJKӋ ӭQJGөQJ[ӱOêҧQKVӱGөQJKӑFPi\YjQKXFҫXFӫDYLӋFQj\WURQJFXӝFVӕQJQJj\ nay. 3KҫQWLӃSWKHRӣ&KѭѫQJQj\FNJQJWUuQKEj\PөFWLrXQJKLrQFӭXFӫDOXұQ YăQ[iFÿӏQKÿӕLWѭӧQJSKҥPYLYjSKѭѫQJSKiSQJKLrQFӭXWURQJVXӕWTXiWUuQK WKӵFKLӋQ&iFQӝLGXQJQj\ÿmÿѭӧFQJKLrQFӭXYjFKӑQOӑFGӵDWUrQWKӵFWӃYjGӵD WUrQFiFNӃWTXҧYӅÿӅFѭѫQJOXұQYăQ &XӕLFQJFҩXWU~FOXұQYăQÿѭӧFÿӅFұSWUuQKEj\VӕOѭӧQJFKѭѫQJQӝLGXQJ WURQJWӯQJFKѭѫQJÿӇQJѭӡLÿӑFKLӇXU}KѫQYӅFiFEѭӟFWKӵFKLӋQYjNӃWTXҧÿѭӧF WUuQKEj\WURQJQJKLrQFӭX 1 - &KѭѫQJ0ӣÿҫX 1 +91JX\ӉQ1JӑF;XkQ+X\ 1ӝLVX\YjWăQJFѭӡQJFKҩWOѭӧQJYLGHRVӱGөQJPi\KӑF &+ѬѪ1* 1.1 GVHD: PGS.TS. Hà Hoàng Kha MӢ ĈҪU Lý do chӑQÿӅ tài +uQKҧQKYjYLGHRPDQJWӟLQKLӅXêQJKƭD YӅYăQKyD ± NKRDKӑF FNJQJQKѭOj QKӳQJ NKRҧQK NKҳF Nӹ QLӋP ÿѭӧF OѭX JLӳ OҥL QKӡ YjR FiF Nƭ WKXұW JKL KuQK 7X\ QKLrQFKҩWOѭӧQJFӫDFiFQӝLGXQJQj\VӁEӏKѭKҥLKRһFPҩWPiWWKHRWKӡLJLDQÿһF ELӋWOjÿӕLYӟLFiFYLGHROӏFKVӱFyWKӡLJLDQOkXÿӡL1JRjLUDFiFÿLӅXNLӋQEҧRTXҧQ QJKqRQjQKRһFGRFKLӃQWUDQKFNJQJFy WKӇ Jk\ҧQKKѭӣQJWӟLYLӋFPҩWPiW FKҩW OѭӧQJYjGӳOLӋXFyWURQJFiFQJXӗQYLGHRQj\+uQKFKRWKҩ\FKLӃQWUDQKNK{QJ FKӍGӯQJOҥLӣFiFPkXWKXүQFKtQKWUӏPjQyFKҥPWӟLPӑLNKtDFҥQKFӫDFXӝFVӕQJ 7URQJÿyWKӇKLӋQYLӋFWUDQKJLjQKFiFFKLӃQOӧLSKҭPJLӳDTXkQĈӗQJPLQK và quân Phát-[tWÿӕLYӟLFiFWiFSKҭPQJKӋWKXұWWURQJFiFEҧRWjQJUҧLUiFӣNKҳSFKkXÆX >@9LӋFQj\ÿһFUDPӝWQKLӋPYөOjFҫQFyFiFF{QJFөÿӇNK{LSKөFFiFJLiWUӏ QJKӋWKXұWEӏPҩWWURQJNKtDFҥQKYӅWUҧLQJKLӋPWKӏJLiFOjUҩWOӟQ Hình 1.1 &KL͇QOͫLSḴPFͯDĈͱFWURQJWK͇FKL͇QWKͱÿ˱ͫFF̭WJLͷW̩L 6FKORVVNLUFKH(OOLQJHQ%DYDULDWKiQJQăP %rQFҥQKÿyKLӋQQD\VӵWLӃQEӝWURQJF{QJQJKӋWK{QJWLQYjWURQJFiFQJjQK F{QJQJKӋFKӃWҥRWKLӃWEӏÿLӋQWӱFiFEӝ[ӱOêKuQKҧQKYLGHRQJj\FjQJÿiSӭQJ FiFQKXFҫXYӅQkQJFDRFKҩWOѭӧQJWK{QJWLQWURQJKuQKҧQKKRһFYLGHRÿy&iFWKLӃW EӏJLDÿuQKEuQKGkQQKѭ79ÿLӋQWKRҥLWK{QJPLQKÿmFyFiFEӝKLӇQWKӏFKҩWOѭӧQJ NKiWӕWĈӕLYӟLFiFYLGHRNK{QJEӏPҩWPiWWK{QJWLQGRKѭKӓQJKRһFGRFKLӃQWUDQK WjQSKiWKuFiFQKXFҫXFkQEҵQJÿѭӧFFKҩWOѭӧQJFӫDFiFYLGHRÿyYӟLFiFEӝKLӇQ WKӏFKҩWOѭӧQJFNJQJNK{QJQJӯQJÿѭӧFÿһFUD+uQKWKӇKLӋQWӕFÿӝSKiWWULӇQÿӝ SKkQJLҧLFӫDPjQKuQKWURQJQăPÿҫXFӫDWKӃNӍYjWDFyWKӇWKҩ\ÿѭӧFWURQJ 1 - &KѭѫQJ0ӣÿҫX 2 +91JX\ӉQ1JӑF;XkQ+X\ 1ӝLVX\YjWăQJFѭӡQJFKҩWOѭӧQJYLGHRVӱGөQJPi\KӑF GVHD: PGS.TS. Hà Hoàng Kha YzQJNKRҧQJPӝWWKұSNӍÿӝSKkQJLҧLPjQKuQKÿmWăQJOrQOҫQGүQWӟLVӵOӛLWKӡL QKDQKFKyQJFӫDFiFYLGHRNK{QJÿiSӭQJNӏSVӵ[XWKӃQj\ +uQK7͙Fÿ͡WăQJWU˱ͥQJÿ͡SKkQJL̫LPjQKuQKWURQJWK̵SQLrQV ĈyOjQKӳQJYҩQÿӅFzQYѭӟQJSKҧLPjQJjQK[ӱOêҧQKSKҧLJLҧLTX\ӃW7X\QKLrQ WURQJQKӳQJQăPJҫQÿk\YLӋFSKiWWULӇQFiFQӅQWҧQJWUҧLQJKLӋPYLGHRFjQJÿѭӧF QJѭӡLGQJYjFiFQKjÿҫXWѭÿӇWkPWӟL+uQKWKӇKLӋQWӕFÿӝWăQJWUѭӣQJQKX FҫX[HPYLGHRWUӵFWX\ӃQWҥLNKXYӵFFKkXȱ 7KiL%uQK'ѭѫQJWURQJJLDLÿRҥQ 2017 ± >@9LӋFQj\OjPFiF{QJOӟQF{QJQJKӋWURQJQJjQKF{QJQJKLӋSYLGHR SKҧLWұSWUXQJKѫQYjRFiFF{QJFөWăQJFѭӡQJFKҩWOѭӧQJFKRFiFYLGHRFӫDPuQK +uQK7͙Fÿ͡WăQJWU˱ͧQJGXQJO˱ͫQJWK͓WU˱ͥQJQJjQKVWUHDPLQJYLGHRNKX Y͹FFKkXȱ 7KiL%uQK'˱˯QJWͳ± 2028 1 - &KѭѫQJ0ӣÿҫX 3 +91JX\ӉQ1JӑF;XkQ+X\ 1ӝLVX\YjWăQJFѭӡQJFKҩWOѭӧQJYLGHRVӱGөQJPi\KӑF GVHD: PGS.TS. Hà Hoàng Kha 9uFiFOêGRWUrQQJѭӡLWKӵFKLӋQOXұQYăQTX\ӃWÿӏQKWKӵFKLӋQÿӅWjLQJKLrQFӭX ³1ӝLVX\YjWăQJFѭӡQJFKҩWOѭӧQJYLGHRVӱGөQJKӑFPi\´0өFWLrXFӫDQJKLrQ FӭXOjVӁÿyQJJySÿѭӧFFKRYLӋFOѭXJLӳYjEҧRWӗQFiFJLiWUӏNKRDKӑFYăQKyD± [mKӝL«WK{QJTXDYLӋFNK{LSKөFFiFYLGHRFyFKҩWOѭӧQJNK{QJFDRKRһFEӏKѭKҥL Ĉk\OjKѭӟQJQJKLrQFӭXÿѭӧFTXDQWkPWURQJFӝQJÿӗQJNKRDKӑFWURQJWKӡLJLDQ JҫQÿk\YjWLӃSWөFOjKѭӟQJQJKLrQFӭXKӭDKҽQWURQJWKӡLJLDQWӟL 1.2 Mөc tiêu nghiên cӭu 0өFWLrXQJKLrQFӭXFӫDÿӅWjLOXұQYăQEDRJӗPFiFÿLӇPFѫEҧQVDX - - - 1.3 1JKLrQFӭu FiFJLҧLSKiSQӝLVX\YjWăQJFѭӡQJFKҩWOѭӧQJYLGHRVӱGөQJ Pi\KӑF7URQJÿyWұSWUXQJYjRQӝLGXQJFKtQK7KrPPjXFKRYLGHRWUҳQJ ÿHQWăQJÿӝSKkQJLҧLYjWăQJIUDPHUDWH 7KDPNKҧRVRViQKQKұQ[pWYӟLFiFF{QJWUuQKQJKLrQFӭXWUѭӟFÿy ĈѭDUD PӝWJLҧLSKiSKӋWKӕQJÿӇFҧLWLӃQKRһFWӕLѭXFKREjLWRiQGӵDWUrQQӅQWҧQJ Pi\KӑF. 7KӵFQJKLӋPFiFJLҧLSKiSKLӋQWKӵFKyDFKѭѫQJWUuQKtrên máy tính Google Colab. ĈѭDUDEҧQJÿiQKJLiNӃWTXҧ ÿmWKӵFKLӋQ ĈӕLWѭӧng và phҥm vi nghiên cӭu 7URQJNKX{QNKәWKӵFKLӋQOXұQYăQKӋWKӕQJWKDPNKҧRFiFQJKLrQFӭXWӯWUѭӟF WKHRÿӏQKKѭӟQJQӅQWҧQJOjKӑFPi\&iFP{KuQKÿѭӧFVӱGөQJWURQJTXiWUuQKWKӵF KLӋQÿӅXÿѭӧFÿӕLFKLӃXNLӇPFKӭQJVRViQKYӟLFiFWjLOLӋXNKRDKӑFÿѭӧFF{QJEӕ EӣLFiFWәFKӭFX\WtQ ĈӇWKӵFKLӋQPөFWLrXQӝLVX\YjWăQJFѭӡQJFKRFiFYLGHRÿѭӧFFKӑQQJѭӡLWKӵF KLӋQVӱGөQJFiFQJXӗQWjLQJX\rQYjSKҫQFӭQJKӧSOêYjKӧSSKiS&iFF{QJFө ÿѭӧF[k\GӵQJGӵDWUrQVӵNӃWKӧSJLӳDWKӯDKѭӣQJFiFNӃWTXҧYjWKӵFKLӋQPӟLFiF SKҫQFҫQWKLӃW &iFNӃWTXҧWKXÿѭӧFWӯTXiWUuQKWKӵFKLӋQOXұQYăQQJRjLJLiWUӏSKөFYөFKR QKӳQJQJKLrQFӭXYjÿiQKJLi VkXKѫQFzQFyWKӇÿѭӧFVӱGөQJÿӇ[k\GӵQJPӝWKӋ WKӕQJNKҧWKLWURQJWKӵFWӃ7X\QKLrQÿӇiSGөQJYjRWKӵFWӃFzQSKөWKXӝFYjR QKLӅX\rXFҫXYӅQӅQWҧQJNӻWKXұWF{QJQJKӋÿLӅXNLӋQWKӡLJLDQÿiSӭQJWjLQJX\rQ ± SKҫQFӭQJYjÿLӅXNLӋQNLQKWӃ- [mKӝL 1.4 3KѭѫQJSKiSQJKLrQFӭu 3KѭѫQJSKiSQJKLrQFӭXÿѭӧFOӵDFKӑQWURQJOXұQYăQOjiSGөQJOêWKX\ӃWÿm ÿѭӧFQJKLrQFӭXNLӇPFKӭQJWURQJFiFWjLOLӋXNKRDKӑFÿѭӧFF{QJEӕJҫQÿk\EӣL FiFWәFKӭFX\WtQ3KѭѫQJSKiSFKӫÿҥROjÿӇWKӯDKѭӣQJKRһFNӃWKӧS[k\GӵQJP{ KuQKJLҧLWKXұWWtQKWRiQYjFiFSKѭѫQJSKiSÿӇQӝLVX\YjWăQJFѭӡQJFKRYLGHR 1 - &KѭѫQJ0ӣÿҫX 4 +91JX\ӉQ1JӑF;XkQ+X\ 1ӝLVX\YjWăQJFѭӡQJFKҩWOѭӧQJYLGHRVӱGөQJPi\KӑF GVHD: PGS.TS. Hà Hoàng Kha .ӃWTXҧVӁÿѭӧFWKӵFKLӋQEҵQJFiFF{QJFөKӧSSKiSFӫD*RRJOH&RODEYjWUrQPi\ WtQKYjVӁÿѭӧFÿiQKJLiWtQKKLӋXTXҧFӫDP{KuQKÿm[k\GӵQJYjSKiWWULӇQ 1.5 Cҩu trúc luұQYăQ 1ӝLGXQJFӫDOXұQYăQEDRJӗPFKѭѫQJ &KѭѫQJOjFKѭѫQJPӣÿҫX&iFQӝLGXQJÿmÿѭӧFWUuQKEj\U}UjQJWURQJFiF SKҫQWUrQ &KѭѫQJFӫDOXұQYăQVӁWUuQKEj\WәQJTXDQYҩQÿӅFKӫÿӅFӫDÿӅWjLÿyOj³1ӝL VX\YjWăQJFѭӡQJFKҩWOѭӧQJYLGHR´YjFiFF{QJWUuQKQJKLrQFӭXOLrQTXDQ &KѭѫQJWUuQKEj\FѫVӣOêWKX\ӃWYӅFiFNKiLQLӋP[ӱOêҧQKVӕYLGHRVӕYjFiF NKiLQLӋPOLrQTXDQ&iFP{KuQKPҥQJQHXUDOQәLWLӃQJYjFyӭQJGөQJWURQJOXұQ YăQFNJQJÿѭӧFJLӟLWKLӋXÿӅFұSPӝWFiFK WәQJTXDQQKҩW &KѭѫQJWUuQKEj\FiFFҩXWU~FP{KuQKOXұQYăQWKӵFKLӋQÿӇQӝLVX\YjWăQJ FѭӡQJFKҩWOѭӧQJFKRYLGHR%rQFҥQKÿyFKѭѫQJQj\FNJQJJLҧLWKtFKU}FiFP{KuQK WKjQKSKҫQÿѭӧFWKӯDKѭӣQJYjӭQJGөQJWKӃQjRWURQJTXiWUuQKWKӵFKLӋQ &KѭѫQJVӁWUuQKEj\TXiWUuQKKLӋQWKӵFFiFP{KuQKWUrQQӅQWҧQJ*RRJOH&RODE FҩXWU~FFӫDFKѭѫQJWUuQKYjFiFF{QJFөKӛWUӧWKӵFKLӋQ&KѭѫQJQj\FNJQJWUuQK Ej\NӃWTXҧÿҥWÿѭӧFWK{QJTXDPӝWEӝYLGHRÿmÿѭӧFQӝLVX\YjWăQJFѭӡQJFKҩW OѭӧQJ7URQJSKҫQQj\OXұQYăQFNJQJÿѭDUDPӝWVӕWK{QJVӕÿӏQKOѭӧQJYjÿӏQK WtQKÿӇÿiQKJLiKLӋXTXҧFӫDTXiWUuQKQӝLVX\YjWăQJFѭӡQJFӫDFiFYLGHRNӃWTXҧ ÿmWKӵFKLӋQÿѭӧF &KѭѫQJVӁNӃWOXұQOҥLWRjQEӝOXұQYăQWyPWҳWNӃWTXҧQJKLrQFӭXÿӗQJWKӡL ÿӅUDKѭӟQJSKiWWULӇQWURQJWѭѫQJODLFӫDOXұQYăQ 1 - &KѭѫQJ0ӣÿҫX 5 +91JX\ӉQ1JӑF;XkQ+X\ 1ӝLVX\YjWăQJFѭӡQJFKҩWOѭӧQJYLGHRVӱGөQJPi\KӑF GVHD: PGS.TS. Hà Hoàng Kha &+ѬѪ1* *,Ӟ,7+,ӊ87Ә1*48$1 &KѭѫQJÿӅFұSWӟLFiFYҩQÿӅWәQJTXDQFӫDÿӅWjLFөWKӇÿyOj³1ӝLVX\Yj WăQJFѭӡQJFKҩWOѭӧQJFKRYLGHRVӱGөQJPi\KӑF´ 3KҫQÿҫXQyLYӅ[XKѭӟQJSKiWWULӇQQKDQKFKyQJFӫDF{QJQJKӋGүQWӟLVӵ WăQJWUѭӣQJUҩWQKDQKYӅFiFFKXҭQFKҩWOѭӧQJFӫDYLGHRYӅÿӝSKkQJLҧLKuQKҧQK ÿӝSKkQJLҧLPjXVҳFFNJQJQKѭWӕFÿӝIUDPHUDWHĈLӅXQj\FNJQJGүQWӟLPӝWKӋTXҧ OjFiFQӝLGXQJKuQKҧQKYLGHRFyWXәLÿӡLWӯ[D[ѭDFyWKӇEӏOӛLWKӡLQKDQKFKyQJ 1KXFҫXFyFiFSKѭѫQJSKiSÿӅWăQJFѭӡQJFiFQӝLGXQJÿyÿѭӧFUDÿӡLĈyFNJQJ OjOêGRÿӅWjL³1ӝLVX\YjWăQJFѭӡQJFKҩWOѭӧQJFKRYLGHRVӱGөQJPi\KӑF´ 3KҫQVDXEjQYӅFiFQJKLrQFӭXWUѭӟFÿk\YӅFiFQӝLGXQJWăQJFѭӡQJFKҩW OѭӧQJYLGHR7URQJÿyÿӅFѭѫQJÿӅ[XҩWYjWұSWUXQJWKҧROXұQӣQӝLGXQJFKtQK FKROXұQYăQĈyOjWăQJFѭӡQJÿӝSKkQJLҧLWKrPPjXFKRYLGHRYjWăQJWӕFÿӝ framerate. 2 ± *LӟLWKLӋXWәQJTXDQ 6 +91JX\ӉQ1JӑF;XkQ+X\ 1ӝLVX\YjWăQJFѭӡQJFKҩWOѭӧQJYLGHRVӱGөQJPi\KӑF &+ѬѪ1* GVHD: PGS.TS. Hà Hoàng Kha GIӞI THIӊU TӘNG QUAN Giҧi thích mӝt sӕ khái niӋm video và các vҩQÿӅ ÿһt ra 2.1 1KѭÿmÿӅFұSWURQJSKҫQWUѭӟFQKXFҫXYӅFҧLWKLӋQFKҩWOѭӧQJYLGHRGӣNKtD FҥQKQKѭNK{LSKөFWK{QJWLQKRһFWăQJFѭӡQJFKҩWOѭӧQJÿӅXOjUҩWWROӟQ7X\QKLrQ ÿӇFyWKӇWKӵFKLӋQÿѭӧFFiFQӝLGXQJÿyPӝWFiFKNKRDKӑFYjEjLEҧQWDSKҧLKLӇX U}YӅFiFWK{QJVӕFҫQÿѭӧF[ӱOêYjFөWKӇFiFWK{QJVӕÿyÿѭӧFTX\ӃWÿӏQKWKӃQjR +m\ÿLӇPTXDQKDQKYӅPӝWVӕWKXұWQJӳ ĈӝSKkQJLҧLWKӇKLӋQTXDWK{QJVӕGjL[UӝQJÿѫQYӏOjSL[HO&yPӝWVӕFKXҭQ YӅÿӝSKkQJLҧLSKәELӃQ í í í í í í 1920 × 1080: Full HD (High Definition) 2048 × 1080: 2K (Digital Cinema) 3840 × 2160: 4K (UHD - Ultra High Definition) 4096 × 2160: 4K (Digital Cinema) 7680 × 4320: 8K (UHD) 15360 × 8640: 16K (UHD) Ngày nay, FKXҭQYLGHRÿӝSKkQJLҧL.QJj\FjQJÿѭӧFVӱGөQJUӝQJUmLYjQJj\ FjQJWLӃQOrQ.7ӯQKӳQJQăP@.tFKWKѭӟFFҧPELӃQFjQJ OӟQWKuÿӝQKҥ\ViQJFjQJWӕWKD\QyLFiFKNKiFOjFKҩWOѭӧQJFӫDPӛLSL[HOVӁFjQJ cao. ĈӕLYӟLJLiWUӏNtFKWKѭӟFSKkQJLҧLFӫDFҧPELӃQQyWKӇKLӋQJLiWUӏÿӝFKLDÿѫQ YӏQKӓQKҩWWUrQFҧPELӃQÿy+DLFҧPELӃQFyFQJNtFKWKѭӟFQKѭQJFyWKӇÿѭӧF FKLDSKkQJLҧLNKiFQKDXĈӅXQj\GүQWӟLNӃWTXҧKuQKҧQKNKiFQKDXҦQKFӫD FҧPELӃQFyÿӝSKkQJLҧLFDRVӁQpWKѫQWX\QKLrQJLiWUӏWK{QJWLQFӫDWӯQJSL[HOVӁ NpP1JѭӧFOҥLҧQKFyÿӝSKkQJLҧLWKҩSKѫQWUrQFQJFҧPELӃQVӁFKRҧQKFyNӃW TXҧNpPVҳFQpWKѫQQKѭQJWK{QJWLQYjFKҩWOѭӧQJFӫDWӯQJSL[HOULrQJOҿVӁFDR KѫQ7\WKXӝFYjRQKXFҫXVӱGөQJPjQKjVҧQ[XҩWVӁOӵDFKӑQFiFFkQEҵQJFiF JLiWUӏVDRFKRSKKӧS 2 ± *LӟLWKLӋXWәQJTXDQ 7 +91JX\ӉQ1JӑF;XkQ+X\