• 検索結果がありません。

Examination for the method of perturbation

N/A
N/A
Protected

Academic year: 2021

シェア "Examination for the method of perturbation"

Copied!
11
0
0

読み込み中.... (全文を見る)

全文

(1)

௉֧໩ɶඬ៟

ᇿᬝʾᅋȣəబࡐ᠔ᡨՀჼ

Examination for the method of perturbation

character recognition using the correlation

ϝ᭢ݜላჼ੘ ׉ೱ༗ໝ ιँᅐᒪ ࡨᅏ᧎߂

Marie Nikaido Kouji Kitamura Yumi Nakashima Michio Yasuda

᣺ඛ

ٛሻɶᇿᬝ໩ʾ΢ȱʭូᇾɈᷳᅖᏩשвɶ௉֧ᇿᬝៜ፡ʾᝑɮȦɀɪɨ௉֧፱܎ʾᏰՒ

׈ʾɸȱʱ᷷ೠᠳబɨɸᷳଚೃȴబࡐ̗͹̻̇͹˺ʾᅋȣə˶͇͓͡͹˶͗ͯࡵᱸʾᝑɮ ȣᷳɕɶ֜ഔʾඬ៟ɍʱ᷷ࡵᱸɰɸᷳᮗᐲቆଚೃȴబࡐ̗͹̻̇͹˺ETL6ɶଚೃȴత ࡐᨃՒʾщᅋɈə᷷ʗəᷳฉ྅̫̇͹ͯɶ࡚ᓀʣʒȽՀჼɪɶᏹʙطʸɑɰɢȣɧʟඬ៟

ɍʱ᷷

1 ̹˴̬ͯ

ɍɨɰݩٗɈə௉֧ᇿᬝ໩[1],[2]ȱʭᷳబࡐ᠔ᡨՀჼɪɶᏹʙطʸɑɰʫʱ֜ഔȳቭ᠔Ʉ ʲɧȣʱ᷷ᇿᬝʾᅋȣə̫̇͹ͯ᠔ᡨɸᷳ௉֧Հჼɪᏹʙطʸɑʱɀɪɰʫʮ̫̇͹ͯᬗɶ˼

͡ʾ᥍ཚɄɑᷳʫʮ޿ȴɮᇿᬝҐʾਁʱثᔥਵȳȞʱ᷷ೠᠳబɨɸᷳ௉֧ᇿᬝ໩ɶвᒛ௉֧ɰ

ဴဠʾᐅʮᷳɕɶ௉֧፱܎ʾᏰՒ׈ɍʱɀɪɨʫʮ᠋ᏰɮᇿᬝҐʾਁʭʲʱثᔥਵɰɢȣɧඬ

៟ʾɍʱ᷷

2 ᇖ୑ಉڐ

ᇿᬝʾ፡ՌɍʱɄȣɰɸᷳ16ూـɶ႒਍૾Ւ[3]ʾᅋȣə᷷బࡐ᠔ᡨɰȮȽʱూـਵ̻˧̘

͟ɸᷳబࡐɶ˺̘͢͹˧૾Ւɰ໡ɠəʟɶɨȞʱȳᷳೠᠳబɰȮȽʱూـਵ̻˧̘͟ɸᷳబࡐ ᅖᏩȱʭᔕಒᅖᏩʋɶ໩ᐸ̻˧̘͟ɪɮɠɧȣʱ᷷16ూـɶ႒਍૾Ւɶ᠋Ᏸɸᷳᑓฑశʝɶ8

ూـɶ̻˧̘͟૾Ւʾᷳᇿࢧ૾Ւɶ೎ဳɰʫʮϝՒɈəʟɶɨȞʱ᷷܏1ɰᷳ႒਍૾Ւɶࡵ᭩

ɶюʾ᝝ɍ᷷܏ɰᅋȣəᅖӆɸᷳࡵᱸɨщᅋɈɧȣʱଚೃȴతࡐȌ0Ȏɶॳܬฉ྅̫̇͹ͯʾ ᅋȣəʟɶɨȞʱ᷷

1

(2)

܏116ూـɶ႒਍૾Ւюdzdzdzdzdzdzdzdzdzdzdzdzdzdzdzdz dzdz (ΨḏԀ૾Ւɶᐲ٫ᷳιḏԏҬɶ૾ՒᷳΩḑ޷Ҭɶ૾Ւ)

→ ր ↑ տ ← ւ ↓ ց

3 ཌྷ჉фЍҸҪ̵ঞᘄ

щᅋɍʱฉ྅̫̇͹ͯɸᷳᰓЭঋʾᅋȣɧщᅋɍʱ˳̷ͯ͟ɶ᧤଼ʾᝑɮɠə[4]᷷ػ΢̫

̇͹ͯᬗɶᰓЭঋʾض˞̕˰͝ɶض˳̷ͯ͟Ƀɪɰ፡ՌɈᷳɕɶᰓЭঋȳᬨҐʾᣰȫɧȣɮȣ ݪطɸᷳฉ྅̫̇͹ͯɪɈɧ஘ᅋɍʱ᷷ʗəೠᠳబɨɸᷳ᧤ՌɈə̫̇͹ͯА޷ɰض˞̕˰͝

ɶॳܬʾɪɠə̗͹̇ʟฉ྅̫̇͹ͯɰهʝʱ᷷

ʗəᷳᬨҐʾᣰȫɧȣɮȣ˳̷ͯ͟А޷ʟ࡚ᓀՀჼʾᝑɮȦɀɪɨᷳฉ྅̫̇͹ͯɰᦝ֐ɍ ʱ᷷ɀɀɨᷳ᧤ՌɄʲəฉ྅̫̇͹ͯʾ࡚ᓀ̫̇͹ͯɪዉɍʱ࡚᷷ᓀ̫̇͹ͯɪщᅋɈəԀ˳

̷ͯ͟ɨ᠔ᡨՀჼʾᝑɮȣᷳ᠞ᠤȳඬՌɄʲəݪطᷳɕɶ˳̷ͯ͟ʾ࡚ᓀ̫̇͹ͯɰᦝ֐ɍ ʱ᷷ɀʲʾ᠞ᠤȳɮȸɮʱʗɨᑲʮȱȫɍɀɪɨ࡚ᷳᓀ̫̇͹ͯʾໃࡱɍʱ᷷

ᰓЭঋ፡Ռɰᅋȣə়ʾ়ᷳ(1)ɰኊɍ᷷

S = (F·G)

||F|| · ||G|| (1)

়(1)ɨɸᷳᅲɮʱ2ɢɶ̫̇͹ͯʾᬝతF,ᬝతGᷳɕɶᰓЭঋʾS ɪɈɧȣʱ᷷ʗ əᷳ||F||ᷳ||G||ɸɕʲɗʲɶᬝతɶᕳॅᇿᬝʾ᝝Ɉᷳ||F||=

F(k,j,i)·F(k,j,i)ᷳ||G||= G(k,j,i)·G(k,j,i)ᷳᇿϠᇿᬝ(F·G)ɸᷳ(F·G) =

F(k,j,i)·G(k,j,i)ɪɍʱ᷷kɸ႒਍ᩁᷳ

j,iɸɕʲɗʲᑓฑూـɶᅖᏩвᒛʾ᝝ɍ᷷

4 ഍۫ፃᱡိ

ೠᠳబɨඬ៟ɍʱ௉֧ᇿᬝ໩ɸᷳвᒛɰᬝɍʱ௉֧ɨȞʱ᷷ɀʲʗɨᅋȣɧȴəвᒛ௉֧

[1],[2],[3]ɸᷳబࡐԀзʾສॳȮʫʁܺᇾɰዊ֧ɄɑʱʟɶɨȞɠə᷷܏2ɰ6×6ᅖᏩɶᎋ שɮ͍̗͟ɨ᝝ɍ᷷ɀɶ͍̗͟ɸతҐȌ2Ȏʾ᝝ɍబࡐɨᷳीҬʾ̫̇͹ͯAᷳدҬʾ̫̇͹

ͯBɪɍʱ᷷ʗəᷳ௉֧፱܎ɸ±1ᅖᏩɪɍʱ̫᷷̇͹ͯAɪ̫̇͹ͯBɶᇿᬝʾ፡Ռɍʱ

(3)

Ʉȣᷳ਀೽ɶвᒛ௉֧ɸ܏2ɶ௉֧̫̇͹ͯB(دΩ)ɶʫȦɰᅖӆԀзʾዊ֧Ʉɑʱ᷷௉֧፱

܎±1ᅖᏩɶ፱܎ɨᷳ௉֧ʾɈɮȣݪطʟهʝɧԀᨃɨ9ɢɶ௉֧̫̇͹ͯʾɪʱ᷷ɀɶ9 ɶ̫̇͹ͯBɪ̫̇͹ͯAɪɶᇿᬝʾɪʮᷳض௉֧Ƀɪɰᐲ٫ʾັʝᷳೋ޿ɪɮʱʟɶʾங Ꮻɍʱ᷷౓ሻɶвᒛ௉֧ʾᅋȣɧೋ޿ᇿᬝҐSʾັʝʱ়ʾ়ᷳ(2)ɰኊɍ᷷ᬝతg,f ɸɕʲ ɗʲ႒਍૾Ւʾهʝʱబࡐᅖӆʾᷳpmaxɸ௉֧ᩁʾ᝝ɍ᷷

S=max

pmax

dy=−pmax

pmax

dx=−pmax

16

k=1

jmax

j=1 imax

i=1

g(k,j,i)·f(k,j+dy,i+dx)

(2)

܏2ḏвᒛɶ௉֧ᇿᬝ໩(਀೽)

̫̇͹ͯA ̫̇͹ͯB (௉֧ձ)

̫̇͹ͯA

̫̇͹ͯB (௉֧ৼ)

← −1ᅖᏩ +1ᅖᏩ

−1↑ ᅖ Ꮹ

+1 ᅖ Ꮹ

΢ూᷳೠᠳబɨඬ៟ɍʱвᒛ௉֧ɸᷳᅖӆԀзɨɸɮȸ᷾ᅖᏩɃɪɰສॳȮʫʁܺᇾɶ௉֧

ʾᝑɮȦ᷷᷾ᅖᏩשвɨɶᇿᬝʾɪʱɀɪɰʫɠɧձᦗɈəଚ໩ʫʮʟೋ޿ᇿᬝɶ᧤Ռ፱܎ȳ Ᏸȱȸɮʱɀɪȱʭᷳʫʮ᠋ᏰɮᇿᬝҐʾਁʱɀɪȳ೚৷ɨȴʱ᷷

(4)

ᠣ౦ɶəʝᷳ܏3ɰ܏2ɪػෲɶ̫̇͹ͯʾኊɍ᷷௉֧፱܎ɸᷳ±1ᅖᏩɪɍʱ᷷܏ʾʙɧ ʟ౦ʭȱɮʫȦɰᷳ௉֧ՀჼʾౄɍݪطᷳࢧዉᅖᏩʾι਒ɰ௉֧ᩁՒɶ᭭஛ᅖᏩȳਓាɪɮ ʱ᷷ʫɠɧᷳࡵ᭩ɰ᧜ᅋɍʱɄȣɸᷳబࡐᅖӆɶԀ፱܎ɰɢȣɧᇿᬝៜ፡ʾᝑɮȦəʝᷳ௉֧

ᩁՒɶо෉᯼ݓʾᅋચɍʱਓាȳȞʱ̫᷷̇͹ͯAɶࢧዉᅖᏩ᷾ᅖᏩ(܏3ḏدᷳ߁ഖᨃՒ)ɪ ɕʲɰࢧਚɍʱ̫̇͹ͯBɶࢧዉᅖᏩȮʫʁ᭭஛8ᅖᏩɶ፱܎(܏3ḏीᷳ1∼9ɰȞəʱ)ɨ ᇿᬝៜ፡ʾᝑɮȣᷳضᅖᏩɃɪɶᐲ٫ʾັʝʱ᷷௉֧ɶᐁഔᷳᐲ٫ȳೋ޿ɪɮɠəʟɶɶʙʾ

஘ᅋɍʱ᷷вᒛ௉֧ʾᅋȣɧೋ޿ᇿᬝҐSʾັʝʱ়ʾ়ᷳ(3)ɰኊɍ᷷

S=

imax−1

i=2 jmax−1

j=

2 max

pmax

dx=−pmax

pmax

dy=−pmax

16

k=

1

g(k,j,i)·g(k,j+dy,i+dx)

(3)

1 2 3 4

5 6

7 8 9

܏31ᅖᏩשвɶвᒛ௉֧ᇿᬝ໩

̫̇͹ͯA ̫̇͹ͯB

5 ᥘ᦬হᶼ

ࡵᱸɰщᅋɈə̗͹̇ɸᷳᮗᐲቆଚೃȴబࡐ̗͹̻̇͹˺ETL6ɶଚೃȴతࡐᨃՒ(0∼9 ɶ10˞̕˰͝)ɨȞʱ̗᷷͹̇ɸ1˞̕˰͝Ȟəʮ1383బࡐȞʮᷳɀʲʾߌతᅮᇼɪҮతᅮ ᇼɶ̗͹̇ɰϝՒɈᷳ1˞̕˰͝Ȟəʮ961బࡐɶϝዘᰓɶ̗͹̇ɪɈɧؙଦȦ(АΩᷳߌత ᅮᇼɶ̗͹̇ʾȌ̫̇͹ͯoȎᷳҮతᅮᇼɶ̗͹̇ʾȌ̫̇͹ͯeȎɪ៥ɍ)᷷ϝՒɍʱɀɪɨ

̗͹̇ɸض691బࡐɪɮʮᷳ1383ᅮᇼɶబࡐɸࡵᱸɨɸщᅋɈɮȣ᷷Ӭ̗͹̇ɸ63×64 Ꮹɶ˨͡͹˺˪͹ᷳ͟ೠࡵᱸɨɸᎋשɮ̥ˑ˼᭎؎ʾౄɈəৼϝҐ׈ʾɈə̗͹̇ʾᅋȣɧȣ ʱ᷷᠔ᡨࡵᱸɸ࡚ᷳᓀ̫̇͹ͯɪೞሻ̫̇͹ͯɶᇿᬝʾᅋȣɧ๪។˞̕˰͝ʾඬᏫɍʱ᷷2 ᰓɶ̗͹̇ʾ࡚ᷳᓀ̫̇͹ͯɪೞሻ̫̇͹ͯɰɕʲɗʲտʮড়ɧᷳ4᦯ʮɶᏹʙطʸɑɨࡵᱸ ʾᝑɮȦ᷷᝝1ɰᷳضᏹʙطʸɑʾ៥ɍ᷷

(5)

᝝1ḏࡵᱸɶᏹʙطʸɑ

ೞሻ̫̇͹ͯ

࡚ᓀ̫̇͹ͯ

̫̇͹ͯo ̫̇͹ͯe

̫̇͹ͯo oo eo

̫̇͹ͯe oe ee

᠔ᡨࡵᱸɰȱȱʱՀჼ౾ᬗሾᑕɶəʝ࡚ᷳᓀ̫̇͹ͯɰщᅋɍʱ̗͹̇ɸᷳȞʭȱɋʝӬ

̗͹̇ȱʭ႒਍ᩁፅʾ፡ՌɈᷳ๪ួ׈ʾౄɈəʟɶʾᅋચɈə᷷

๪ួ׈˳ˑ˼ɸᷳబࡐ᯼ݓȳ8×8ᅖᏩᷳо෉᯼ݓȳ10×10ᅖᏩɪɈᷳо෉᯼ݓɶι਒ɰ బࡐ᯼ݓʾᨔᒛɍʱ᷷௉֧ᇿᬝ໩ɰ᧜ᅋɈə௉֧፱܎ɸᷳ±1ᅖᏩɪɈə᷷

ʗəೠᠳబɨɸᷳʒȽՀჼʾౄɈəݪطɶࡵᱸʟᝑɮɠə᷷ʒȽՀჼɰ᧜ᅋɈəᨿʙѢతʾ

᝝2ɰᷳɕɈɧᷳࡵᱸɰᅋȣəᄦނʾ᝝3ɰ៥ɍ᷷᝝2ɶ|h|ɸສॳూـɶᷳ|v|ɸܺᇾూـɶ ࢧዉᅖᏩȱʭɶ᣿ᮍʾ᝝ɍ᷷

᝝2ḏᨿʙѢత

0 1

0 52 21

1 21 7

᝝3ḏࡵᱸᄦނ

MPU Intel(R) Xeon(R) CPU 5160@ 3.00GHz ԏᨃ៥૞᝾ᒛᅋᩁ 2GB

ˮ̫ͯˑ͛ Intel(R) C++ Compiler Version 10.0

|v|

|h|

6 হᶼᕅ๘

6.1 ᵗձ૏

ೠᠳబɨඬ៟ɍʱ௉֧Հჼʾؙʮӿʲəࡵᱸɶᐁഔ࡚ᷳᓀ̫̇͹ͯɪೞሻ̫̇͹ͯɶᰓЭঋ ȳ100%ʾᣰȫɧɈʗȦݪطȳȞɠə᷷܏4ɰᷳ௉֧೎ʮ·࡚ᓀ̫̇͹ͯо૾ɶᬨҐ95%А ΨɪɈəݪطɶᷳᰓЭঋɶ̭˺̘˨͉͛ʾ᝝ɍ᷷ɀɶ˨̳͛ɸᷳ6910బࡐԀɧɶ˳̷ͯ͟ɶ

๪។˞̕˰͝ᰓЭঋʾ˞˔̘ͯɈəʟɶɨȞʱ᷷ࡵᐸȳ௉֧Հჼɶʙɶᐁഔᷳ቎ᐸȳ௉֧Հჼ ɶЅɰʒȽՀჼʾౄɈəᐁഔɪɮʱ᷷ᑓ᥊ȳ˳̷ͯ͟తᷳฑ᥊ȳᰓЭঋʾ᝝Ɉᷳ100%ʾ1000 ɪɍʱ᷷

܏4ɶ˨̳͛ȱʭᷳoooeeoeeԀɧɶᏹʙطʸɑɰȮȣɧᷳᰓЭঋȳ1000ʾᣰȫʱݪ طȳȞʱɀɪȳቭ᠔ɨȴʱ᷷ʒȽဳɈɶݪطǴᰓЭঋɸೋ޿ɨ1400ᦌȸɰɮɠəʟɶʟȞɠ ə᷷ᰓЭঋɶ̰͹˧ɸᷳɭɶᏹʙطʸɑʟ1000ձৼɪɮɠə᷷

ೠ೽ᷳᰓЭঋɸػ΢̫̇͹ͯɨȞʱɪ100%ɰɮʱɪȣȦʟɶɨȞʱȳᷳӱᦗɶɪȮʮೠࡵ

ᱸɨਁʭʲəᰓЭঋȱʭɸ100%ʾᣰȫʱݪطȳቭ᠔Ʉʲɧȣʱ᷷ɀʲɸᷳೠᠳబɨᅋȣəв ᒛ௉֧ɪ๪ួ׈Հჼɪɶᏹʙطʸɑɰʫʱ৬ᯮɨȞʱɪᓏȫʭʲʱ᷷

ᇿᬝៜ፡ɰщᅋɈɧȣʱబࡐ̗͹̇ɸᷳ63×64ᅖᏩɶ̗͹̇ȱʭబࡐ፱܎ʾՓʮՌɈᷳɕ ɶబࡐ፱܎ɶᨃՒɶʙʾ8×8ᅖᏩɶ̗͹̇ʋɪ๪ួ׈ɈəʟɶɨȞʱ᷷ʫɠɧᷳ๪ួ׈ɶɄ

(6)

ȣɶᑕࣃɸᷳض˳̷ͯ͟Ƀɪɰޱ׈ɍʱɀɪɰɮʱ᷷

ʗə๪ួ׈ՀჼɶɄȣᷳӬᅖӆɪ๪ួ׈ৼɶᅖӆɶᅖᏩػޣȳ᷾ࢧ᷾ɨࢧਚɍʱɪȣȦɀɪ ɸʐʒᣮɀʭɮȣ᷷ࡵ᭩ɰɸᷳࢧਚɍʱঌฉвᒛɶᦌ஛ḃᅖᏩɰᨔՒɍʱɀɪɪɮʱ᷷ᦌ஛ḃ ᅖᏩʋɶᨔՒɶտط(Ѣత)ɸᷳɕɶᐲ٫ȳضᅖᏩɃɪɰ᷾ɪɮʱʫȦɰៜ፡Ɉɧȣʱ᷷

ೠᠳబɨᅋȣəвᒛ௉֧ɸᷳ᷾ᅖᏩɃɪɰࢧዉᅖᏩɪ᭭஛ᅖᏩʾȞʸɑəḎᅖᏩɪɶᇿᬝʾ ɪʮᷳɕɶҐȳೋ޿ɪɮʱʟɶʾ஘ᅋɈɧȣʱ᷷ḎᅖᏩɶɭʲʾ஘ᅋɍʱȱɸضᅖᏩɶៜ፡ᐁ ഔɰʫɠɧޱ׈Ɉܑᷳࡱɨɸɮȣ᷷ɢʗʮᷳ๪ួ׈ɶɄȣɰೋ޿ɨ᷾ɪɮʱʫȦտʮড়ɧʭʲ ɧȣʱѢతɶᨔՒȳɏʲɧȸʱɀɪɰɮʱ᷷Ѕ˳̷ͯ͟ɪɶЪᑕɶຘოʟɕʴɠɧȣɮȣə ʝᷳᅲɮʱ˳̷ͯ͟ᬗɨɸ႒ɰᷳ๪ួ׈ɰʫʱ˼͡ɶ৬ᯮʾؚȽɧɈʗȦثᔥਵȳȞʱ᷷

ɀɶ๪ួ׈ɰʫʱ˼͡ʾᑀ٫ɍʱࢧፎɪɈɧᷳʒȽՀჼʾౄɈəࡵᱸʾᝑɮɠə᷷܏4ȱ ʭᷳʒȽဳɈɶݪطɪຘʍɧᰓЭঋȳԀзᇒɰཚࢶɈɧȣʱɀɪȱʭɕɶ֜ഔȳቭ᠔ɨȴʱ᷷

ʒȽဳɈɶݪطɶᰓЭঋɸᷳೋ޿ɨ1400ᦌȸɶҐȳቭ᠔ɄʲəȳᷳʒȽ೎ʮɶݪطɸೋ޿ɨ ʟ1200ᦌȸʗɨɰཚࢶɈɧȣʱ᷷

0 500 1000 1500 2000 2500 3000

700 800 900 1000 1100 1200 1300 1400 normal with blur

0 500 1000 1500 2000 2500 3000

700 800 900 1000 1100 1200 1300 1400 normal with blur

0 500 1000 1500 2000 2500 3000

700 800 900 1000 1100 1200 1300 1400 normal with blur

0 500 1000 1500 2000 2500 3000

700 800 900 1000 1100 1200 1300 1400 normal with blur

܏4ḏ᠔ᡨ౾ɶᰓЭঋ̭˺̘˨͉͛

oo oe

eo ee

܏5ɰ܏4ɶ̭˺̘˨͉͛ʾ˞̕˰͝ɃɪɰՒȽɧ᝝ኊɈəʟɶʾኊɍ᷷˨̳͛ɰщᅋɈ ə̗͹̇ɸᷳʒȽՀჼဳɈɨᏹʙطʸɑȳooɶʟɶɨȞʱ᷷

˨̳͛ȱʭᷳԀ˞̕˰͝ɰȮȣɧᰓЭঋ1000ձৼɰ̰͹˧ȳቭ᠔ɨȴʱ᷷ʗəᷳ˞̕˰

͝ ”ɶݪطɶʙᰓЭঋȳ1300∼1400ውঋʗɨቭ᠔ɨȴʱ᷷ɀʲɸᷳ᷾ɪȣȦబࡐᅖӆɶ

(7)

๪ួ׈ՀჼɰᬝѢɈɧȣʱɪਰʸʲʱ᷷Ԋзᇒɰɸᷳ᷾ɪȣȦబࡐɸᷳЅɶబࡐɪຘ᥏Ɉɧᑓ ᬌɶৣႢʾɈɧȮʮᷳ๪ួ׈ৼʟᑓᬌɶৣႢʾѰɧʱʫȦᷳສॳూـɶЪᑕოʾងదɈɧȣʱ ɪȣȦဠɨȞʱ᷷

0 50 100 150 200 250 300 350

700 800 900 1000 1100 1200 1300 1400 normal 0oo

0 50 100 150 200 250 300 350

700 800 900 1000 1100 1200 1300 1400 normal 1oo

0 50 100 150 200 250 300 350

700 800 900 1000 1100 1200 1300 1400 normal 2oo

0 50 100 150 200 250 300 350

700 800 900 1000 1100 1200 1300 1400 normal 3oo

0 50 100 150 200 250 300 350

700 800 900 1000 1100 1200 1300 1400 normal 4oo

0 50 100 150 200 250 300 350

700 800 900 1000 1100 1200 1300 1400 normal 5oo

0 50 100 150 200 250 300 350

700 800 900 1000 1100 1200 1300 1400 normal 6oo

0 50 100 150 200 250 300 350

700 800 900 1000 1100 1200 1300 1400 normal 7oo

0 50 100 150 200 250 300 350

700 800 900 1000 1100 1200 1300 1400 normal 8oo

0 50 100 150 200 250 300 350

700 800 900 1000 1100 1200 1300 1400 normal 9oo

܏5ḏض˞̕˰͝Ƀɪɶ᠔ᡨ౾ɶᰓЭঋ̭˺̘˨͉͛(oo)

category ”0” category ”1” category ”2”

category ”3” category ”4” category ”5”

category ”6” category ”7” category ”8”

category ”9”

6.2 ঞᘄфЍҸҪ̵ϰҪїғ൨̤ᥢᥨ൰ঔ൨

щᅋɍʱᇿᬝៜ፡(௉֧Հჼʾهʚ)ɪᬨҐ(3፯؏၄)ɰʫʮ࡚ᷳᓀ̫̇͹ͯɪɈɧඬՌɄ ʲʱ˳̷ͯ͟తɸޱ֧ɍʱ᷷܏6ɪ᝝4ɰ௉֧ᇿᬝʾᅋȣə࡚ᓀ̫̇͹ͯɶ᧤଼తʾኊɍ࡚᷷

ᓀ̫̇͹ͯɰ஘ᅋɄʲਁʱԀ˳̷ͯ͟తɸᷳॳܬ̫̇͹ͯʟطʸɑɧ6920బࡐɨȞʱ᷷˨͛

̳ɶᑓ᥊ɸ᧤଼˳̷ͯ͟తᷳฑ᥊ɸᬨҐʾ᝝ɍ᷷

ʗɏɸᷳвᒛ௉֧೎ʮɶݪطɰɢȣɧᦗʍʱ᷷܏6ᷳ᝝4ȱʭᷳҮతɪߌతɰʫʱ᧐ȣɸత

(8)

బࡐ∼כబࡐውঋɪɮʮᷳʒȽ೎ʮ·ဳɈɶؕ

ూɨ̫̇͹ͯɰʫʱॄɰ႒ࡱɶӀـɸʙʭʲɮ ȱɠə᷷ʒȽՀჼɰᬝɈɧ໳ᇼʾɈɧʙʱɪᷳ

ᬨҐ97%ʾނɰʒȽ೎ʮɶݪطɶ᧤଼˳̷ͯ͟

తȱʭ1%Ƀɪɰ1000బࡐውঋɶΨౢȳʙʭ ʲᷳʒȽဳɈɶݪطɰຘʍ᧤଼˳̷ͯ͟తɶΨ

ౢȳᘼɈȣɀɪȳቭ᠔ɨȴʱ᷷ɀʲɸᷳӱɰᦗ ʍəᰓЭঋɶ̭˺̘˨͉͛ȱʭᠣ౦ɨȴʱ᷷ʒ ȽဳɈɰຘʍɧᷳʒȽ೎ʮɶూȳԀзᇒɰᰓЭ ঋȳгȣ᷷ʫɠɧᷳʒȽ೎ʮɶݪطɶూȳ͝˺

̘ɰᦝ֐ɍʱɄȣɰັʝʭʲʱᬨҐȳгȸɮʱ ɪȣȦɀɪɰɮʱ᷷

0 1000 2000 3000 4000 5000

90 91 92 93 94 95 96 97 98 99 pettern o

pettern e pettern o with blur pettern e with blur

܏6ḏ᧤଼࡚ᓀ˳̷ͯ͟త (௉֧೎ʮ)

๓ɰᷳ௉֧ᇿᬝ໩ʾщᅋɈɮȱɠəݪطɶ࡚ᓀ̫̇͹ͯɶ˳̷ͯ͟తɪຘ᥏ʾɍʱ᷷᝝5 ɰᷳ௉֧ဳɈɶݪطɶ࡚ᓀ̫̇͹ͯɶ˳̷ͯ͟తʾኊɍ᷷௉֧ᇿᬝʾщᅋɈɧȣɮȣəʝೋ޿

ɶᰓЭঋɸ100%ɪɮʮᷳັʝʭʲʱᬨҐɶҐɸ޿ȴȸᅲɮɠɧȸʱ᷷ɕɶəʝᷳ᧤଼˳̷ͯ

͟తȳػውঋɶ፱܎ɨȞʱ̗͹̇ʾ଻Ꭺɍʱ᷷

᝝5ȱʭ̫ᷳ̇͹ͯeɪoɶॄȳᬨҐ88%АΨɨ100బࡐАΨඬՌɄʲɧȣʱɀɪȳቭ᠔ ɨȴʱ᷷௉֧೎ʮɶݪطɪຘʍɧᷳԀзᇒɰ̫̇͹ͯᬗɶॄȳ޿ȴȸɮʱӀـȳʙʭʲʱ᷷

᝝4ḏ᧤଼࡚ᓀ˳̷ͯ͟త(௉֧೎ʮ)

90% 91% 92% 93% 94% 95% 96% 97% 98% 99%

o 116బࡐ 136బࡐ 177బࡐ 239బࡐ 358బࡐ 600బࡐ 1035బࡐ 1643బࡐ 2480బࡐ 3520బࡐ

e 93 113 134 200 329 616 1010 1664 2530 3500

o 109 133 151 193 278 437 812 1567 2864 4465

e 99 99 111 158 255 438 808 1509 2778 4319

ʒȽ Հჼ ̫̇

͹ͯ

ᬨҐ

Ɉ

ʮ

᝝5ḏ᧤଼࡚ᓀ˳̷ͯ͟త(௉֧ဳɈ)

80% 82% 84% 86% 88% 90% 92%

o 258బࡐ 357బࡐ 500బࡐ 758బࡐ 1275బࡐ 2185బࡐ 3602బࡐ

e 223 303 443 680 1159 2028 3469

̫̇

͹ͯ

ᬨҐ

๓ɰᷳ᠔ᡨ˶͇͓͡͹˶͗ͯɶᐁഔ(᠞ᠤబࡐతȮʫʁ๪ᠤო)ɰɢȣɧɶᐁഔʾ៥ɍ᷷܏7 ɰ௉֧ᇿᬝ໩ʾᅋȣəݪطɶ᠞ᠤబࡐతɶ˨̳͛ʾᷳ᝝6ɰ᠞ᠤబࡐతɪ๪ᠤოʾኊɍ᷷ػɋ

̫̇͹ͯɶᏹʙطʸɑ(oo, ee)ɰɢȣɧɸ࡚ᷳᓀ˳̷ͯ͟ʾ᧤଼ɍʱ౾ဠɨ᠞ᠤȳဳȸɮʱʗ ɨՀჼʾᑲʮȱȫɈɧȣʱəʝᷳᐁഔʾሂᅪɈɧȣʱ᷷

(9)

ᬨҐȳ޿ȴȣɪщᅋɍʱ࡚ᓀ˳̷ͯ͟తȳޅȫʱəʝᷳড়ဵ᠔ᡨɶᏁঋɸΨȳʱӀـɰȞ ʱ᷷əɚɈᷳຘ᥏ɍʱࢧዉȳޅȫʱɀɪɨʫȽȣɮ᠞ᠤʾޅʣɍثᔥਵʟɨɧȸʱəʝᷳਓɏ ɈʟᬨҐȳ޿ȴȣʐɭ᠔ᡨოȳᖧȸɮʱʸȽɨɸɮȣ(юɪɈɧᷳʒȽՀჼဳɈ ᏹʙطʸɑoe ɶ94%∼96%ɶݪطፅ)᷷ʒȽՀჼɶ೎ဳɨຘ᥏ʾɍʱɪᷳ܏7ȱʭʟʸȱʱʫȦɰᷳʒȽ೎

ʮɶݪطɶూȳᏹʙطʸɑɰʫʱ᠞ᠤబࡐతɶॄȳࢶɮȣӀـȳȞʱɀɪȳቭ᠔ɨȴʱ᷷ɀɶ ɀɪȱʭᷳʒȽՀჼɸᰓЭঋɶងదɚȽɨɸɮȸ̫ᷳ̇͹ͯɶ᧐ȣɰʫʱॄʾ᥍ཚɄɑʱ֜ഔ ʟȞʱɪᓏȫʭʲʱ᷷

௉֧ᇿᬝ໩ʾᅋȣəೋᖧɶᐁഔɪɈɧɸᷳᬨҐ99%ᏹʙطʸɑoeɶݪطɨᷳ᠞ᠤబࡐత 10బࡐ(๪ᠤო99.86%)ɪɮɠə᷷ʒȽՀჼ೎ʮɶݪطɰᬝɈɧʟػෲɶᐁഔɨȞʮ̫ᷳ̇͹

ͯɰʫʱॄʾ᥍ཚɍʱɀɪɸɨȴɧʟ᠔ᡨᏁঋʾΨȾʱȣȦဠɨɸ႒ፃɍʍȴ֜ഔɸਁʭʲɮ ȱɠə᷷

0 5 10 15 20 25 30 35

93 94 95 96 97 98 99

oeeo

0 5 10 15 20 25 30 35

93 94 95 96 97 98 99

oe with blur eo with blur

܏7ḏ᠞ᠤబࡐత(ीḏʒȽဳɈᷳدḏʒȽ೎ʮ)

᝝6ḏ᠞ᠤబࡐతɪ๪ᠤო(௉֧೎ʮ)

ᬨҐ 93% 94% 95% 96% 97% 98% 99%

oe 15బࡐ 10బࡐ 11బࡐ 12బࡐ 11బࡐ 10బࡐ 10బࡐ Ɉ eo 26 24 17 16 15 15 15

oe 20 22 19 20 14 12 10 ʮ eo 19 21 18 13 11 12 11

oe 99.78% 99.86% 99.84% 99.83% 99.84% 99.86% 99.86% Ɉ eo 99.62 99.65 99.75 99.77 99.78 99.78 99.78

oe 99.71 99.68 99.73 99.71 99.80 99.83 99.86 ʮ eo 99.73 99.70 99.74 99.81 99.84 99.83 99.84 ʒȽ

Հჼ ᏹʙطʸɑ

๓ɰᷳ௉֧ᇿᬝ໩ʾᅋȣɮȱɠəݪطɶࡵᱸᐁഔʾ᝝7ɰ៥ɍ᷷ӱɰʟᦗʍəʫȦɰᷳᬨҐ ɶ፱܎ɸ௉֧ᇿᬝ໩ʾᅋȣəݪطɪ޿ȴȸᅲɮʱəʝᷳ᧤଼࡚ᓀ˳̷ͯ͟తɶᐁഔʟᓏૂɈɮ ȳʭຘ᥏ɍʱ᷷

(10)

ೋᖧɶᐁഔɨᷳᬨҐ92%ᏹʙطʸɑoeɶݪطɶ᠞ᠤబࡐత5బࡐ(๪ᠤო99.93%)ɨȞɠ ə᷷᧤଼࡚ᓀ˳̷ͯ͟తɸ3602బࡐ(̫̇͹ͯo)ɨȞʮᷳ௉֧ᇿᬝ໩ʾᅋȣəݪطɨػፅɶ

᧤଼࡚ᓀ˳̷ͯ͟తʾ୨ɢ̗͹̇ɸᷳᬨҐ99%ɶ3520బࡐ(ʒȽဳɈ ̫̇͹ͯo)ɪɮʱ᷷

ɀɶ2ɢɶ̗͹̇ʾຘ᥏ɍʱɪᷳ᠞ᠤబࡐత5బࡐᷳ๪ᠤოɰɈɧ0.07%௉֧ဳɈɶూȳΨ ʗʸɠɧȣʱ᷷᧤଼˳̷ͯ͟తʾᓏૂɈɧЅɶᬨҐɶݪطʾຘ᥏Ɉɧʟᷳʐʒ௉֧ဳɈɶݪط ɶ๪ᠤოɶూȳΨʗʸʱᐁഔɪɮɠə᷷

᝝7ḏ᠞ᠤబࡐతɪ๪ᠤო(௉֧ဳɈ)

ᬨҐ 80% 82% 84% 86% 88% 90% 92%

oe 18బࡐ 12బࡐ 14బࡐ 12బࡐ 8బࡐ 6బࡐ 5బࡐ

eo 23 19 17 13 10 9 9

oe 99.74% 99.83% 99.80% 99.83% 99.88% 99.91% 99.93% eo 99.67 99.73 99.75 99.81 99.86 99.87 99.87 ᏹʙطʸɑ

᠞ ᠤ బࡐత

6.3 ڄቀෂᱛ

ೋৼɰᷳೠᠳబɨᝑɮɠə᠔ᡨ˶͇͓͡͹˶͗ͯࡵᱸɶՀჼ౾ᬗɰɢȣɧݩٗɍʱ᷷܏8ɪ

᝝8ɰ᠔ᡨՀჼɰាɈəՀჼ౾ᬗʾኊɍ᷷˨̳͛ɸ௉֧ᇿᬝ໩ʾщᅋɈəʟɶɨᷳᏹʙطʸɑ ɸoeɶ̗͹̇ɨȞʱ᷷ฑ᥊ɸᬨҐ(שвḏ%)ᷳᑓ᥊ɸՀჼ౾ᬗ(שвḏኾ)ɪɍʱ᷷ʗəᷳࡵ

ᐸɸʒȽՀჼဳɈɶݪطᷳ቎ᐸɸʒȽՀჼ೎ʮɶݪطʾ᝝ʸɍ᷷

0 50 100 150 200 250 300 350

92 93 94 95 96 97 98 99

normal with blur

܏8ḏ᠔ᡨՀჼ౾ᬗ(вᒛ௉֧ᇿᬝщᅋᷳᏹʙطʸɑoe)

᝝8ḏ᠔ᡨՀჼ౾ᬗ(вᒛ௉֧ᇿᬝщᅋ)

92% 93% 94% 95% 96% 97% 98% 99%

oe 11s 19s 27s 44s 75s 119s 185s 267s

Ɉ eo 11 16 25 45 73 121 190 272

oe 13 15 21 32 59 114 214 333 ʮ eo 10 13 20 32 58 109 206 322 ʒȽ

Հჼ ᏹʙطʸɑ ᬨҐ

(11)

᠔ᡨՀჼ౾ᬗʾඬ៟ɍʱΨɨᷳᨿាɰɮɠɧȸʱɶɸ࡚ᓀ̫̇͹ͯɶబࡐత(4؏၄)ɨ Ȟʱ᷷᠔ᡨՀჼɨɸᷳ᧤ՌɄʲə࡚ᓀ˳̷ͯ͟Ԁɧɪೞሻబࡐ̫̇͹ͯɪᇿᬝʾɪʱəʝ࡚ᷳ

ᓀ̫̇͹ͯȳ޺ȣውՀჼ౾ᬗʾាɍʱɀɪɰɮʱ᷷

᝝4ȱʭᬨҐ97%АΩɶ̗͹̇ɨɸᷳʒȽՀჼဳɈɶూȳ࡚ᓀ˳̷ͯ͟తȳ޺ȣ᷷ɀɶဠ ʾᤎʗȫɧ܏8ʣ᝝8ɶՀჼ౾ᬗʾඬ៟ɍʱɪᷳʒȽՀჼʫʮʟοɰ࡚ᓀ̫̇͹ͯɶబࡐతȳ Հჼ౾ᬗɰ৬ᯮʾȮʫʒɈɧȣʱɀɪȳቭ᠔ɨȴʱ᷷ᏹʙطʸɑoeᬨҐ92%ɶݪطɨɸᷳʒ ȽՀჼ೎ʮɶూȳ2ኾ޺ȸȱȱɠɧȣʱȳᷳɕʲА޷ɶ̗͹̇ɨɸᷳᬨҐ98%ʾᣰȫʱʗɨ Հჼ౾ᬗȳᦢ᥇ɍʱɀɪɸဳȣ᷷ɀɶɀɪȱʭᷳ᠔ᡨՀჼɰȮȣɧʒȽՀჼɰʫʱՀჼ౾ᬗɶ ৬ᯮɸᷳʐʒဳȣʟɶɪਰʸʲʱ᷷

7 ̤ͥͯ

ࡵᱸɶᐁഔȱʭᷳᓏ൙ɍʱвᒛ௉֧ᇿᬝ໩ȱʭɸ႒ፃɍʍȴ֜ഔɸቭ᠔ɨȴɮȱɠə᷷Ɉȱ ɈᷳᰓЭঋɶڦᰊɮɭೠᠳబɶ˶͇͓͡͹˶͗ͯࡵᱸɨɸʗɚ។ໃɈɧȣɮȣϜ᯲ȳȞʱɀɪ ȱʭᷳఏᖧɶлҐɸȞʱɪਰʸʲʱ᷷

ϾৼɶᠦᰊɪɈɧᷳᰓЭঋɶងదʾهʝə๪ួ׈ՀჼɶូᇾɈʣЅɶ௉֧Հჼɪɶᏹʙطʸ ɑፅᷳɄʭɮʱඬ៟ȳਓាɨȞʱ᷷

ݓᘓ൰ሊ

[1] ׉ೱᷳϝ᭢ݜᷳιँᷳࡨᅏḏ”᧜ਚ௉֧ᇿᬝ໩ɶ஭൙ᷳѲ࡚ବݩᷳPRMU2005-77 [2] ׉ೱᷳϝ᭢ݜᷳιँᷳࡨᅏḏ”బࡐ᠔ᡨɶəʝɶ᧜ਚ௉֧ᇿᬝ໩ɶ஭൙”ᷳ౦౭޿࡚ቆዲ

Ꮨា–੸ݩ࡚ᨃ14رᷳॳ૾183

[3] ϝ᭢ݜᷳ׉ೱᷳιँᷳࡨᅏḏ”ᐸᏩ႒਍ɶଡ଼৐ɪɕɶ֜ഔ”ᷳ౦౭޿࡚ቆዲᏘា੸ݩ࡚ᨃ– ጽ19ر ॳ૾233

[4] ϝ᭢ݜᷳ׉ೱᷳιँᷳࡨᅏḏ”ᰓЭঋᝑ՛ʾᅋȣʱ࡚ᓀ˳̷ͯ͟ɶ᧤଼໩”ᷳ౦౭޿࡚ቆ ዲᏘា–੸ݩ࡚ᨃ20ر3

参照

関連したドキュメント

Assume that F maps positive definite matrices either into positive definite matrices or into negative definite matrices, the general nonlinear matrix equation X A ∗ FXA Q was

Based on the Perron complement P(A=A[ ]) and generalized Perron comple- ment P t (A=A[ ]) of a nonnegative irreducible matrix A, we derive a simple and practical method that

It is suggested by our method that most of the quadratic algebras for all St¨ ackel equivalence classes of 3D second order quantum superintegrable systems on conformally flat

Homotopy perturbation method HPM and boundary element method BEM for calculating the exact and numerical solutions of Poisson equation with appropriate boundary and initial

Next, we prove bounds for the dimensions of p-adic MLV-spaces in Section 3, assuming results in Section 4, and make a conjecture about a special element in the motivic Galois group

Transirico, “Second order elliptic equations in weighted Sobolev spaces on unbounded domains,” Rendiconti della Accademia Nazionale delle Scienze detta dei XL.. Memorie di

In this paper, we establish some iterative methods for solving real and complex zeroes of nonlinear equations by using the modified homotopy perturbation method which is mainly due

The numerical tests that we have done showed significant gain in computing time of this method in comparison with the usual Galerkin method and kept a comparable precision to this