1. Trang chủ
  2. » Thể loại khác

kết hợp các bộ phận phân lớp SVM cho việc nhận dạng chữ việt viết tay rời r

10 2 0

Đang tải... (xem toàn văn)

THÔNG TIN TÀI LIỆU

Thông tin cơ bản

Tiêu đề Kết hợp Các Bộ Phận Phân Lớp SVM cho Việc Nhận Dạng Chữ Việt Viết Tay Rời
Tác giả Phạm Anh Phương, Ngô Quốc Tào, Lương Chi Mai
Trường học Khoa Công Nghệ Thông Tin, Trường Đại Học Huế
Chuyên ngành Xử lý ảnh và Học máy
Thể loại Đề án tốt nghiệp
Năm xuất bản 2009
Thành phố Huế
Định dạng
Số trang 10
Dung lượng 245,27 KB

Các công cụ chuyển đổi và chỉnh sửa cho tài liệu này

Nội dung

Bai bao nghien ciiu mot sd loai dac trung co the ap dung cho bai toan nhan dang cbd Viet viet tay rdi rac.. Tir do de xuat mot mo hinb nhan dang chii Viet viet tay rdi rac tren co sd phu

Trang 1

Tap ciii' Tin hpc va Bleu khiin hpc, T.25, S.l (2009), 88-97

KET Hgp CAC BO PHAN PHAN iCfP SVM

CHO VIEC NHAN DANG CHU" VIET VIET TAY R6^I RAC

PHAM ANH P H U 0 N G \ NGO QUOC TAO^, LUONG CHI MAI^

^ Khoa Cong nghe Thong tin, Truang Dgi hpc Khoa hgc Hue

^ Viin Cong nghi thong tin, Viin Khoa hgc vd Cong nghi Viet Nam

A b s t r a c t This paper studies some features, which can be applied to Vietnamese handwritten character recognition Base on SVM classification and Haar wavelet features we propose a new model for Vietnamese handwritten recognition Our test results over Vietnamese handwriting with 50,000 character samples show the relatively high accuracy of our recognition model

T d m t a t Bai bao nghien ciiu mot sd loai dac trung co the ap dung cho bai toan nhan dang cbd Viet viet tay rdi rac Tir do de xuat mot mo hinb nhan dang chii Viet viet tay rdi rac tren co sd phuang phap vec t o t u a ket hop vdi lua chon dac trung wave-let Haar Cac ket qua thuc nghiem tren cac tap dii lieu chii viet tay tieng Viet vdi 50000 mau t u thu thap cho thay mo binh nhan dang

de xuat dat do chfnh xac tuong ddi cao

1 G l C r i T H I E U Nhan dang chir viet t a y dang la van de thach t h u c ldn ddi vdi c i e n h a nghien ciiu Cho den nay, bai t o i n nhan dang chur viet tay van chua cd d u g c mgt giii p h a p t o n g t h e Mdt

sd ket q u i chd yeu chi t a p trung tren cac t a p dii lieu chir sd viet tay chuan n h u USPS va MNIST 14, 5, 6], ben canh dd cung cd mdt sd edng trinh nghien ciiu tren c i c he ehfr cii La tinh, Hy Lap, Trung Qudc t u y nhien c i c ket q u i cung chi d u g c gidi ban trong mot p h a m

vi hep 12, 7] D i e biet ddi vdi viec nhan dang chii viet tay tieng Viet lai eang gap nhieu khd khan hon do bo ky t y tieng Viet ed nhieu ky t y vdi hinh dang r a t gidng nhau, chi k h i c nhau chut ft ve phan d a u Do dd r a t ft cdng trinh nghien ciiu ve n h i n dang chii viet tay tieng

Viet Bai toan chiing tdi d a t ra 6 day la xay d y n g mdt md hinh nhan d a n g chii Viet in viet

tay rdi rac Bd ky t y tieng Viet bao gdm t i p ky t y khdng d a u A, B, C, D, D, E, G, H, I, K,

L, M, N, O, P, Q, R, S, T, U, V, X, Y va cac ky t y cd d a u A, A, A, A, A, A, A, A, A, A, A,

A, A, A, A, A, A, E, E, E, E, E, E, E, E, E, E, E, I, I, 1, I, I, 6, O, 6, 6, 6, 6, 0, 6, 6,

6, 6, 0, a, a, o, 6, o, u; u, u, u, u, y, ir, u; u, u, u, Y, Y, Y, Y, Y chung tdi gidi

ban pham vi cda bai t o i n theo mgt sd qui dinh nhu: c i c chir viet p h i i cd mot k h o i n g c i c h

t u o n g ddi, giu'a phan chur va phan dau p h i i tach rdi nhau

P h u a n g p h a p vec t o t y a (SVM Support Vector Machines) la mdt p h u a n g p h a p m i y hoc tien tien d a cd nhieu t h a n h cdng khdng chi trong eae linh vyc khai p h i dtr lieu m a cdn trong linh vyc nhan dang Trong nhurng t h i p nien gan day, SVM d a d u g c i p dung rgng rai v i o

Trang 2

nhieu bai t o i n t h y c te va cho nhieu ket q u i rat k h i quan | 1 , 3] Vi vay, ehdng tdi sU dung

phuang p h i p vec t o t y a cho mo hinh nhan dang ma chung tdi de xuat TVong mdt he thdng

nhan dang thi viec trfch chgn d i e trung l i mot b u d c quan trgng, nd cd i n h hudng rat ldn

den do chfnh xac cda he thdng n h i n dang Cd rat nhieu p h u a n g p h i p trich chgn d i e trung

hieu q u i ed t h e i p dung cho chur viet tay nhu: ma tran trgng sd, t o i n td' Kirsch, cac bieu dd

chieu 14, 5, 7], trong bai b i o nay chung tdi da sdr dung v i cii dat t h d nghiem tren mgt sd

cic loai dac t r u n g dd va quyet dinh sd dung y t u d n g cda p h u a n g p h i p trfch chgn dac trung

wavelet Haar l8] cho mo hinh n h i n dang chii Viet viet tay rdi rac

Tiep theo, Mue 2 se tdm t a t nhung y t u d n g ca b i n cda p h u o n g phap vec t o tua Mue

3 trinh bay cac ket q u i t h y c nghiem tren dii lieu chu' viet tay tieng Viet vdi mdt sd p h u a n g

p h i p trfch chgn dac t r u n g thdng dung Mue 4 p h i t hga kien true cda md hinh nhan dang

chir Viet viet tay rdi rac va c i c ket q u i t h y c nghiem theo mo hinh nay Cudi cung la phan

ket luan va h u d n g phat trien

2 P H U O N G P H A P V E C T O TU*A

Cho t i p mau huan luyen xi £ R^, i = 1, , N vk cac nhan t u a n g ung yi e { - 1 , +1},

mue tieu cda SVM la tim mgt sieu phang phan each (dugc xac dinh bdi w) sao cho khoing

each (margin) giiia hai Idp d a t cyc dai (Hinh 1)

Hinh 1 Sieu phang tach vdi khoing cich cyc dai

Ham mue tieu cda mgt m i y phan ldp SVM nhi p h i n cd the dugc phat bieu nhu sau:

gix) = w.^ix)-\-b, (1)

trong do, vec t a dau vao x G R^, w la vec t o chuan cda sieu phang phan cich trong khdng

gian dac t r u n g d u g e sinh ra tur i n h xa cda bam $(a;) : R^ -^ P ^ ( M > TV, $(a;) cd the

tuyen tfnh hoac phi tuyen) va b Ii do lech so vdi gdc tga do ll], SVM gdc dugc thiet ke cho

bai toan p h i n Idp nhi p h i n , vi vay dau cda gix) cho biet vec t o x thugc ldp +1 hay ldp —1

Viec tim sieu p h a n g phan eich chi'nh l i viec giii bai t o i n qui hoach toan phuang

(QP-Quadratic programming):

1 7,

m a x ( a i — -a Ha) (2)

a 2

Trang 3

90

thda man 0 < Oi <C, ii vk ^ j ^ i Qiyi = 0, ' trong dd a = ]ai, ,a/v]-^ va H la ma tran kfeh thude N x N dugc ggi la ma trin nhan

(kernel matrix) vdi mdi phan tir Hii,j) = yiyj^ixi).^ixj)

Giii bai toin QP (3) ta thu dugc:

"^aiVi^ixi) (3)

Mdi mau huan luyen Xi tuong iing vdi mdt he sd Lagrange cnj Cic miu cd cii > 0 dugc

ggi la vec to hd trg

The (3) vao (1), ta cd:

9ix) = Y2aiyi^ixi).^ix) -b 6 (4)

Gii sir ^ixi).^ixj) = Kixi,Xj) Nghia la, tfch vd hudng trong khdng gian die trung

tuong duong vdi mgt him nhin K cda khdng gian dau via Vi viy, ta khdng can phii tfnh

tryc tiep cac gia tri $(xi),$(xj) ma chi can tfnh tfch vd hudng < $(a;j).$(a;^) > giin tiep

thdng qua ham nhan Kixi, Xj) Nhu vay, ta ed ham myc tieu cho bai toan phan ldp SVM ed

gix) = ^ atyiKixi, x) -\- b (5)

3 N H A N D A N G C H U V I E T TAY VOTI C A C D A C T R U N G T H O N G D U N G

TVong phan nay, ehung tdi gidi thieu mgt sd die trung da dugc sii dung rat hieu qui trong

cic bai toin nhan dang chir viet tay 14, 5, 7]

3.1 Trong so viing (Zoning)

m

Hinh 2: Trfch chgn die trung trgng sd viing

Anh ky ty dugc chia thanh N x N vung (zones) Tong sd diem den cda moi vimg se duoc

chgn de tao thanh vec to die trung

TVong thye nghiem, vdi inh kich thudc 16 x 16, chung tdi chon N = 8, nhu vav cd

8 X 8 = 64 die trung

Trang 4

3.2 B i ^ u d o c h i e u ( P r o j e c t i o n h i s t o g r a m s )

S'^' %

Hinh 3 Trfch chgn c i c bieu dd chieu ngang, dge va 2 d u d n g cheo

Y t u d n g ca b i n cda p h u a n g p h i p trfch chgn dac t r u n g nay l i chieu c i c diem den tren

inh 2 chieu theo c i c h u d n g ngang, dge v i hai d u d n g cheo thanh mgt day c i c tin hieu 1 chieu

U'u diem cda c i c dac t r u n g nay l i khdng phu thugc vao nhieu, tuy nhien no van phu thudc

vao do nghieng cda chur

TVong t h y c nghiem, vdi i n h kfeh thudc 16 x 16, chung tdi chgn 16 ngang -bl6 dge -1-2 x 16

cheo = 64 d i e t r u n g

3.3 TVich c h o n c h u t u y e n ( C o n t o u r profiles)

rAJ

Hinh 4 TVi'ch chgn cac khdi ben ngoii cda chii

Phan d u g c trich chgn l i khoing cich tii bien cda khung chiia i n h tdi diem den dau tien

cda chii tren cung 1 ddng quet P h u a n g p h i p trfch chgn n i y md t i tdt cac khdi ben ngoii

cda chir va cho phep phan biet mdt sd lugng ldn c i c ky t y

TVong t h y c nghiem, vdi i n h ki'ch thudc 16 x 16, cd 16 t r i i -|-16 phii -1-16 tren -|-16 dudi

= 64 dac trung

3.4 Trfch c h g n d a c t r u t i g wavelet Haar

Chung tdi sd dung y t u d n g cda p h u a n g p h i p trfch chon dac trung wavelet Haar |8] de

chgn t i p d i e t r u n g cho mdi i n h ky t y dau vao

Tu i n h nhi phan kich t h u d c 2n(2n (Hinh 5), q u i trinh tn'ch chon d i e trung dugc md t i

theo thuat t o i n sau:

Procedure HaarFeature

Input Ma tran vudng (A, n) cap 2"

Output T a p c i e d i e trirng { F i , F2, P2"x2'' }•

Trang 5

92

0 0 0

O D D

0 0 0

0 0 0

0 0 0

0 0 0

0 0 0

O OD

O 0 0

0 0 1

O 0 1

o i l

1 1 1

1 1 0

o o o o

0 0 DO oooo

0 0 0 1

0 0 1 1

0 1 1 1

1 1 1 1

1 1 O Q

1 1 0 0

1 0 0 0 oooo oooo

0 0 1 1

D i l i

1 1 1 1

1 1 0 1

1 1 0 0

1 0 0 0

1 1 1 1 oooo oooo oooo

1 0 0 0 0

1 O O D 0

1 0 0 0 0

1 0 0 0 0

1 1 0 0 0

1 1 O D 0

1 1 1 0 0

1 1 1 0 0

1 1 1 1 0

0 0 1 1 1

0 0 0 1 1

0 0 0 1 1

0 0 0 1 1

0 0 0 0 0 0 0 0

D D O D D O O D

O O D O ' l O D I

O D D l^-^O 1 1

0 0 0 0 0 0 1 1

0 O D O O 1 1 1

0 0 0 0 1 1 1 1

0 0 0 1 1 1 1 1

0 0 1 1 1 0 0 0

D O I CJ^i:! D 0

o i l ^^3 0 0

1 1 1 0 0 0 0 0

1 1 1 0 0 0 0 0

0 1 1 1 0 0 0 c

1 1 1 1 O D O C

1 1 1 1 O D O C

1 1 1 c-ioo

1 0 1 l ? A o 0

1 0 D 1 1 0 0 C

0 0 0 1 1 1 D C

0 0 0 0 1 1 0 c 1 1 1 1 1 1 0 c

0 D 0 D 0 1 1 C

o 0 0 0 4 1 1 1

G 0 O'-'^O 1 1

0 0 0 0 0 0 1 1

0 0 0 0 0 0 1 1

Hinh 5 Ttn'ch chgn die trung wavelet Haar

Method

1 Khdi tao: Queue = $; z = 1;

2 - Tfnh Fi bang tdng eie diem den trong toin bd ma tran (A, n);

- PUSH{iA,n),Queue);

3 While Queue ^ # Do

{

- POPiQueue,iA,n));

- if (n > 1)

{

Chia inh thanh 4 phan: Ai, A2, A3, A4;

for (i = l;i = 4;i-b-b)

PUSHi(Aj,ndiv2), Queue);

}

- Ggi 5 1 , 52, 53, 54 la tdng cic diem den tuang urng vdi Ai, A2, A3, A4;

- Tfnh Fi+i = S1 + 52;

Pi+2 = 5 2 - b 5 3 ; F; + 3 = 54;

-i = i-\-3;

}

Trong thyc nghiem, vdi phan chii chung tdi chgn n = 4, nhu viy ta cd: 1 -b 3 4- 4 x 3 -t

4 x 4 x 3 - b 4 x 4 x 4 x 3 = 256 die trung, edn vdi phan dau chung tdi chon n = 3, nhu viy

cd tat c i 64 dac trung

Phuang phap trich chgn die trung nay se tao ra mgt day sd cic dac trung giim dan, Vdi ciing mdt chur thi cic gii tri ldn d dau day tuong ddi on dinh, cd the dai dien cho hinh dang

khii quit cua chu; cdn cac gia tri 6 cudi day nhd dan va khdng dn dinh, the hien sy da dang

trong tung chi tiet cda chii

3.5 K e t qud thu'c nghiem

Dii lieu chii viet tay tieng Viet cda chung tdi dugc thu thap tir 655 ngtrdi viet khic nhau ddi tugng chd yeu la sinh vien Mdi ngudi viet khoing 200 chu' in hoa, cac ky ty duac viet

Trang 6

KET HOP CAC Bp PHAN PHAN LOP SVM 9 3 rdi rac Chung tdi chgn Igc ra 50000 mau de tien h i n h t h y c nghiem, trong dd sd dung 30000 mau dilng de huan luyen va 20000 man phuc vu cho viec nhan dang

Cdng cu phan ldp d u g c x i y dyng t u cac SVM nhi phan Chung tdi chgn thuat t o i n SMO ]1] de huan luyen c i c m i y phan Idp nhi p h i n theo chien luge mot chdng mgt (OVO- One verus One) vdi c i c t h a m sd C = 100 va sdr dung ham nhan la ham Gaussian vdi c = 0, 5

Bdng 1 Ket q u i n h i n dang theo c i c loai d i e t r u n g k h i c nhau

T a p d i e trung Zoning Projection histograms Contour profiles wavelet Haar

Do chi'nh x i c 75.37%

73.82%

73.14%

77.28%

Cac ket q u i t h y c nghiem d b i n g 1 cho t h a y ddi vdi b i i t o i n nhan dang chir viet tay tieng Viet, i p dung d i e t r u n g Haar cho do chfnh xac cao han so vdi sii dung c i c d i e trung khac Tuy nhien, do bg ky t y tieng Viet qua da dang nen viec i p dung c i c d i e trung nay len t a p dii lieu viet tay tieng Viet van chua dat hieu q u i cao Vi v i y can p h i i cd mgt md hinh hieu

q u i cho bai t o i n nhan dang chii viet tay tieng Viet TVong phan sau chung tdi se de xuat mgt mo hinh hieu q u i cho viec nhan dang chir viet tay tieng Viet

4 M O H I N H N H A N D A N G CHLT V I E T V I E T T A Y RClI R A C

[ ^ ( ( 3 ^ ) ^ Trich ch^ti

dac tnrng ^^

Anh

16x16

y

Ket qua n h ^ d ^ s

Hinh 6 Kien true cda he nhan dang chu viet tay tieng Viet

TVong phan nay, chung tdi se trinh bay chi tiet kien true cda md hinh n h i n dang chu Viet viet tay rdi rac Tren c a sd c i c t h i n h phan lien thdng cda i n h , chung tdi phan t i p ky t y tieng Viet thanh 3 nhdm va t i c h c i c ky t y cd dau thanh c i c phan rdi nhau Sau dd chung toi

Trang 7

94

xiy dyng cac may phan ldp SVM de nhan dang eho ttrng phan chir va dau, cudi cung ghep ndi cic ket qui phan ldp cda cic phan chir va phan dau de cd dugc ket q u i nhin dang cuoi cung

4.1 Tien xiV ly

Myc dfch cda giai doan tien xii ly nham ting do chfnh xac cda he thdng nhan dang Khi quet inh thudng gap cic Ioai nhieu, vi vay chung toi sdr dung mdt sd ky thuat lgc nhieu de khdr cic nhieu ddm va nhieu vet dai Ddi vdi nhieu ddm, sdr dung cic bd Igc trung binh va Igc trung vi, cdn vdi cic nhieu vet dai thi chung tdi sd dung phuang phip khii cic vimg lien thdng nhd (Hinh 7)

De thuin tien cho viec xii Iy sau, nay, chiing tdi bien doi inh dau vao tir inh da cap xim thanh inh nhi phan

O /

A A I

(a) Nhieu ddm (b) Nhieu vet dai

Hinh 7 Mdt sd nhieu thudng gap khi quet inh

Chuan hda inh theo vung lien thdng Chuan hda inh nham myc dich tao dieu kien thuin tien cho cdng doan tich inh thinh ti^rng phan chir va dau

Bu"6"c 1: Xic dinh eie vung lien thdng tren inh (Hinh 8)

" -1 -2

0-(a) (b)

Hinh 8 Chuan hda inh

(a) Anh gdc

(b) Xae dinh cic viing lien thdng va dinh thu ty cic vung lien thdng

Du'o'c 2: Sip xep cic vung lien thdng theo thii ty tir tren xudng (Hinh 8b)

Btfo-c 3:

0

m

(a) ^ (b) (e)

Hinh 9 Chuan hda eie vung lien thdng

Trang 8

- Neu i n h chi cd 1 vimg lien thong: Chuan hda i n h ve ki'ch thudc chuan 16 x 16 (Hinh

9a)

- Neu i n h cd 2 vung lien thdng: Ggi 5(z) la dien tfch vung lien thdng t h d i

Neu 5(1) > 5 ( 2 ) thi dau cda phan lien thdng 2 la dau nang (.) va chl can chuan hda

vdng lien thdng 1 ve kich t h u d c chuan 16 x 16

Ngugc lai: T i c h i n h t h i n h 2 phan: phan chu va phan dau Chuan hda phan chu ve kich

thudc chuan 16 x 16 va phan dau ve ki'ch t h u d c chuan 8 x 8 (Hinh 9b),

- Neu i n h cd 3 vung lien thdng:

Neu 5(3) = m l n { 5 ( i ) } thi dau cda phan lien thong nay la dau nang (.) Do dd chi can

chuan hda thanh phan lien thdng 1 ve kfeh t h u d c chuan 8 x 8 v i thanh phan lien thdng 2 ve

kich thudc chuan 16 x 16,

Ngugc Iai: T i c h i n h thanh 3 phan t u cic vung lien thdng Chuan hda cac viing lien thong

1 va 2 ve kich t h u d c chuan 8 x 8 va chuan hda vung lien thong 3 ve kich thudc chuan 16 x 16

(Hinh 9c)

4.2 P h a n n h o m so" b o

Dya vao sd t h a n h phan lien thdng chung tdi t i c h bg ky t y tieng Viet thanh 3 nhom:

Nhdm 1: Nhdm cd 1 vung lien thdng A, B, C, D, D, E, G, H, 1, K, L, M, N, O, P, Q R, S, T,

U, V, X, Y, O, U'

Nhdm 2: Nhdm cd 2 vimg lien thong A, A, A, A, A, A, A, E, E, E, E, E, E, I, t, I, I, I, 6 , 6 ,

6,0,6, o, 6, 6, 6,6, o, u, u, u, u, u, u; u, u', u, y, Y, Y, Y, Y, Y,

Nhom 3: Nhdm cd 3 vung lien thdng A, A, A, A, A, A, A, A, A, A, E, E, E, E, E, 6, 6 , 6 ,

4.3 Trich c h g n d a c t r u t i g

Chung tdi sii dung d i e t r u n g wavelet Haar, Vdi phan chd, chung tdi chgn kich thudc i n h

16 x 16, nhu v i y t a c d : l - t - 3 - b 4 x 3 i - 4 x 4 x 3 - | - 4 x 4 x 4 x 3 = 256 d i e trung, cdn vdi

phan dau kfeh t h u d c i n h 8 x 8 thi cd t a t c i 64 d i e trung

4.4 X a y drfng c a c m a y p h a n l d p S V M

TVong phan niy, chung tdi se xay dyng 3 may p h i n ldp SVM, sd dung t a p d i e trung

wavelet Haar de huan luyen phan Idp va nhan dang,

SVMl: p h i n Idp ddi vdi nhdm ky t y cd 1 vilng lien thdng {A, B, C, D, D, E, G, H, I K L,

M, N, O, P, Q, R, S, T, U, V, X, Y, O, U'}

SVM2: ddi vdi c i c ky t y cd dau thi phan chir deu la cic nguyen am vi vay may nay chi phan

ldp cac nguyen am {A, E, I, O, U, Y}

SVM3: phan Idp cac dau {/, \ , ?, ~ , A, V} (sac, huyen, hdi, nga, dau 5, dau a)

Cac m i y p h i n ldp S V M l , SVM2, SVM3 dugc x i y dyng t u cic SVM nhi p h i n Chung

tdi chgn t h u a t t o i n SMO |1] de huan luyen cic m i y p h i n Idp nhi p h i n theo chien luge mgt

chdng mdt (OVO One verus One) vdi c i c tham sd C = 100 va su dung ham nhan la ham

Gaussian vdi CT = 0, 5

Trang 9

96

4.5 Ket qud thu'c nghiem

Vdi 50000 mau ma chiing tdi thu thap de tien hanh thyc nghiem, trong dd 13782 mau chu khdng dau dung de huan luyen, phan cdn lai phuc vu cho viec nhan dang

Ba tap dii lieu dugc xay dyng phuc vu cho viee huan luyen:

• TrainDatal: Tap cic dau tieng Viet {/, \ , ?, ~, A, V}, vdi 2485 mau

• TrainData2: Tap cic chir cii nguyen im tieng Viet {A, E, I, O, U, Y}, vdi 4128 mau

• TVainDataS: Tap cac chir cii tieng Viet khdng dau {A, B, C, D, D, E, G, H, 1, K, L, M, N,

O, P, Q, R, S, T, U, V, X, Y, O, U'}, vdi 13782 mau

f

A

§"

i

i

^

1

A

£

E

6

/

^

E

0

0

A

X

E

6

6

A

A

£

5

L)

I

d

A

£

C

i(

A-A

E

0

0

A

/T

E

6~

0

A

A

E

6

i

Hinh 10 Cic mau trfch tir tap ky ty viet tay tieng Viet

Bdn tap dii lieu dugc xay dyng phuc vu cho viec nhan dang, ket q u i thyc nghiem the hien d Bing 1:

• TestData 1

• TestData 2

• TestData 3

• TestData 4

Tap cac ky ty tieng Viet cd 1 vung lien thdng, vdi 7143 mau

Tap eac ky t y tieng Viet cd 2 vung lien thdng, vdi 16856 mau

Tap cac ky ty tieng Viet cd 3 vimg lien thdng, vdi 12219 mau

= TestData 1 U TestData 2 U TestData 3

Bdng 2 Ket qui nhan dang tren cic t i p dii lieu tieng Viet viet tay rdi rac

T i p mau TestData 1 TestData 2 TestData 3 TestData 4

Sd miu

7143

16856

12219

36218

Do chi'nh xac 82.24%

90.69%

87.78%

88.04%

Cac ket qui thuc nghiem d Bing 2 cho thay md hinh nhan dang chfi Viet viet tay rdi rac

ma chung toi de xuat dat do ehinh xic cao ban rat nhieu so vdi cic ket qui thuc nghiem d Bing 1

5 KET L U A N

Bai bio da sii dung mgt sd phuang phip trfch chgn dac trung de ap dung vao bai toan nhan dang chii viet tay rdi rac Cie ket qui thyc nghiem tren dii lieu chii viet tay tieno- Viet

cho thay i p dung 6.ac trung wavelet Haar cho do chfnh xac cao nhat trong sd cic die trung

dugc lya chgn Vi viy chung tdi da de xuat md hinh nhan dang chtr Viet viet tay rdi rac dya tren co sd phuang phap vec t a tya ket hgp vdi trfch chgn dac trung wavelet Haar Cic

Trang 10

ket q u i t h y c nghiem cda chung tdi tren md hmh nhan dang chii Viet viet tay rdi rac dat do

chi'nh rat k h i quan TVong t u a n g lai, ehung toi tiep tuc nghien cuu phan hau xU ly, i p dung

md hinh ngon ngii N-Gram de c i i thien do chfnh xac cda md hinh nhan dang

T A I L I E U T H A M K H A O

II] J Piatt, Fast training of support vector machines using sequential minimal optimiza-tion,

Ad-vences in Kernel Methods - Support Vector Learning, Cam-bridge, M.A, MIT Press, 1999

(185-208)

12] Le Hoai Bac, Le Hoang Thai, Neural network & genetic algorithm in application to handwritten

character recognition, Tgp chi Tin hpc vd Di'iu khien hpc 1 7 (4) (2001) 57-65

13] Chih-Cbung Chang and Chil-Jen Lin, "LIBSVM: a Library for Support Vector Ma-chines",

National Taiwan University, 2004

14] D Gorgevik, D Cakmakov, An efficient three-stage classifier for handwritten digit recognition

Proceedings of 17^^ Int Conference on Pattern Recognition, 1CPR2004, Vol 4, IEEE

Com-puter Society, Cambridge, UK, 23-26 August 2004 (507-510)

15] D Cakmakov, D Gorgevik, Handwritten digit recognition using classifier coopera-tion schemes

Proceedings of the ^ ' ^ Balkan Conference in Informatics, BCI 2005; Ohrid, November

17-19, 2005 (23-30)

16] Pham Anh Phuang, Nhan dang chii viet tay ban che voi mo binh SVM, Tgp chi khoa hpc Dgi

hpc Hue (42) (2007) 157-165

|7] G Vamvakas, B Gates, 1 Pratikakis, N Stamatopoulos, A, Roniotis, and S, 1, Perantonis,

Hybrid off-line OCR for isolated handwritten greek characters The Fourth lASTED

Interna-tional Conference on Signal Processing, Pattern Recognition, and Applications (SPPRA

2007) ISBN: 978-0-88986-646-1, Innsbruck, Austria, February 2007 (197-202)

18] P Viola, M Jones, Rapid object detection using a boosted cascade of simple features, Proc

Intl Conf on Computer Vision and Pattern Recognition (CVPR) 1, Kauai, HI, USA, 2001

(511-518)

Nhgn bdi ngdy 3 - 3 - 2008 Nhdn lai sau siia ngdy 6 - 2 - 2009

Ngày đăng: 08/12/2022, 21:16

Nguồn tham khảo

Tài liệu tham khảo Loại Chi tiết
12] Le Hoai Bac, Le Hoang Thai, Neural network &amp; genetic algorithm in application to handwritten character recognition, Tgp chi Tin hpc vd Di'iu khien hpc 1 7 (4) (2001) 57-65 Sách, tạp chí
Tiêu đề: Neural network & genetic algorithm in application to handwritten character recognition
Tác giả: Le Hoai Bac, Le Hoang Thai
Năm: 2001
13] Chih-Cbung Chang and Chil-Jen Lin, "LIBSVM: a Library for Support Vector Ma-chines", National Taiwan University, 2004 Sách, tạp chí
Tiêu đề: LIBSVM: A Library for Support Vector Machines
Tác giả: Chih-Cbung Chang, Chil-Jen Lin
Nhà XB: National Taiwan University
Năm: 2004
14] D. Gorgevik, D. Cakmakov, An efficient three-stage classifier for handwritten digit recognition. Proceedings of 17^^ Int. Conference on Pattern Recognition, 1CPR2004, Vol. 4, IEEE Com- puter Society, Cambridge, UK, 23-26 August 2004 (507-510) Sách, tạp chí
Tiêu đề: Proceedings of the 17th International Conference on Pattern Recognition
Tác giả: D. Gorgevik, D. Cakmakov
Nhà XB: IEEE Computer Society
Năm: 2004
15] D. Cakmakov, D. Gorgevik, Handwritten digit recognition using classifier coopera-tion schemes. Proceedings of the ^ ' ^ Balkan Conference in Informatics, BCI 2005; Ohrid, November 17-19, 2005 (23-30) Sách, tạp chí
Tiêu đề: Handwritten digit recognition using classifier cooperation schemes
Tác giả: D. Cakmakov, D. Gorgevik
Nhà XB: Proceedings of the Balkan Conference in Informatics, BCI 2005
Năm: 2005
18] P. Viola, M. Jones, Rapid object detection using a boosted cascade of simple features, Proc. Intl. Conf. on Computer Vision and Pattern Recognition (CVPR) 1, Kauai, HI, USA, 2001 (511-518).Nhgn bdi ngdy 3 - 3 - 2008 Nhdn lai sau siia ngdy 6 - 2 - 2009 Sách, tạp chí
Tiêu đề: Rapid object detection using a boosted cascade of simple features
Tác giả: P. Viola, M. Jones
Năm: 2001
16] Pham Anh Phuang, Nhan dang chii viet tay ban che voi mo binh SVM, Tgp chi khoa hpc Dgi hpc Hue (42) (2007) 157-165.|7] G. Vamvakas, B. Gates, 1. Pratikakis, N. Stamatopoulos, A, Roniotis, and S, .1, Perantonis, Hybrid off-line OCR for isolated handwritten greek characters. The Fourth lASTED Interna- tional Conference on Signal Processing, Pattern Recognition, and Applications (SPPRA 2007) ISBN: 978-0-88986-646-1, Innsbruck, Austria, February 2007 (197-202) Khác

TÀI LIỆU CÙNG NGƯỜI DÙNG

TÀI LIỆU LIÊN QUAN

🧩 Sản phẩm bạn có thể quan tâm

w