• ๋Œ€ํ•œ์ „๊ธฐํ•™ํšŒ
Mobile QR Code QR CODE : The Transactions of the Korean Institute of Electrical Engineers
  • COPE
  • kcse
  • ํ•œ๊ตญ๊ณผํ•™๊ธฐ์ˆ ๋‹จ์ฒด์ด์—ฐํ•ฉํšŒ
  • ํ•œ๊ตญํ•™์ˆ ์ง€์ธ์šฉ์ƒ‰์ธ
  • Scopus
  • crossref
  • orcid

  1. (Dept. of Computer Science, Sangmyung University, Korea.)



image compression, integral imaging, H.264/AVC, elemental images, view image array

1. ์„œ ๋ก 

3์ฐจ์› ์ž…์ฒด ์˜์ƒ์ด ์ฃผ๋ชฉ๋ฐ›๊ธฐ ์‹œ์ž‘ํ•˜๋ฉด์„œ 3์ฐจ์› ๋ฌผ์ฒด๋ฅผ ์ทจ๋“ํ•˜๊ณ  ๋””์Šคํ”Œ๋ ˆ์ดํ•˜๊ธฐ ์œ„ํ•œ ๋‹ค์–‘ํ•œ ๋ฐฉ๋ฒ•์ด ์—ฐ๊ตฌ๋˜์–ด ์™”๋‹ค[1]-[5]. ํŠนํžˆ, ์ง‘์  ์˜์ƒ (integral imaging) ๋ฐฉ์‹์€ ์ž์—ฐ๊ด‘์„ ์ด์šฉํ•˜์—ฌ 3์ฐจ์› ๋ฌผ์ฒด๋ฅผ ๊ธฐ๋กํ•˜๊ณ  ์žฌ์ƒํ•  ์ˆ˜ ์žˆ๋Š” ๊ธฐ์ˆ ๋กœ์„œ 1908๋…„์— Lippmann์— ์˜ํ•ด์„œ ์ฒ˜์Œ ์ œ์•ˆ๋œ ์ดํ›„ ๋งŽ์€ ์—ฐ๊ตฌ๊ฐ€ ์ˆ˜ํ–‰๋˜์–ด ์™”๋‹ค[6]-[12]. ์ง‘์  ์˜์ƒ ๊ธฐ์ˆ ์€ ๊ธฐ์กด์˜ ์ž˜ ์•Œ๋ ค์ง„ ์Šคํ…Œ๋ ˆ์˜ค ์Šค์ฝ”ํ”ผ ๋ฐฉ์‹๊ณผ๋Š” ๋‹ฌ๋ฆฌ ๊ด€์ฐฐ์ž์˜ ์‹œ๊ฐ์  ํ”ผ๋กœ๊ฐ์ด ์—†๊ณ  ์—ฐ์†์ ์ธ ์‹œ์ ๊ณผ ์ˆ˜ํ‰, ์ˆ˜์ง ์‹œ์ฐจ๋ฅผ ๋ชจ๋‘ ์ œ๊ณตํ•  ์ˆ˜ ์žˆ๋‹ค.

๋ Œ์ฆˆ ๋ฐฐ์—ด์„ ํ†ตํ•ด ์ทจ๋“๋œ ์˜์ƒ์„ ์š”์†Œ ์˜์ƒ ๋ฐฐ์—ด(EIA)์ด๋ผ๊ณ  ํ•œ๋‹ค. EIA์˜ ๊ฐ ์š”์†Œ ์˜์ƒ ํ•˜๋‚˜ํ•˜๋‚˜๋Š” 3์ฐจ์› ๋ฌผ์ฒด์˜ ์‹œ์ ์— ๋Œ€ํ•œ ์ •๋ณด๊ฐ€ ๋‹ด๊ฒจ์žˆ๊ณ  ์ด๋Š” ์ผ๋ฐ˜ ์˜์ƒ๊ณผ๋Š” ๋‹ค๋ฅธ ํŠน์ง•์ ์ด๋‹ค. ํ•˜์ง€๋งŒ ์ผ๋ฐ˜์ ์ธ ์˜์ƒ๊ณผ ๋‹ฌ๋ฆฌ ๋งŽ์€ ์‹œ์  ์˜์ƒ์„ ์ €์žฅํ•ด์•ผ ํ•ด์„œ ๋ฐ์ดํ„ฐ์–‘์ด ๋งค์šฐ ๋งŽ๋‹ค๋Š” ๋ฌธ์ œ์ ์ด ์žˆ๋‹ค. ๋”ฐ๋ผ์„œ ์š”์†Œ ์˜์ƒ ๋ฐฐ์—ด์„ ์••์ถ•ํ•˜๋Š” ๋ฌธ์ œ๊ฐ€ ์ง‘์  ์˜์ƒ ๋ถ„์•ผ์—์„œ ์š”๊ตฌ๋œ๋‹ค. ์ด๋Ÿฌํ•œ ์ด์œ ๋กœ ์ง‘์  ์˜์ƒ์— ๋Œ€ํ•œ ์••์ถ• ์—ฐ๊ตฌ๊ฐ€ ํ™œ๋ฐœ์ด ์ด๋ฃจ์–ด์ ธ ์™”๋Š”๋ฐ M. C. Forman et al.์€ ๋‹จ๋ฐฉํ–ฅ ์ง‘์  ์˜์ƒ์— ๋Œ€ํ•ด 3์ฐจ์› DCT(3D-DCT)์— ๊ธฐ๋ฐ˜ํ•˜์—ฌ ์••์ถ• ๊ธฐ๋ฒ•์„ ์ œ์•ˆํ•˜์˜€๋‹ค[13]. R.Zaharia๋Š” 3D-DCT์— ๊ธฐ๋ฐ˜ํ•œ ์ ์‘์  ์–‘์žํ™” ๊ธฐ๋ฒ•์„ ์ œ์•ˆํ•˜์˜€๋‹ค[14]. DCT ์ด์™ธ์—๋„ KLT(Karhune-Loeve Transform)๋ฅผ ์ด์šฉํ•œ ์ง‘์  ์˜์ƒ ์••์ถ• ๋ฐฉ๋ฒ•์ด ์ œ์•ˆ๋˜์—ˆ๋‹ค[15]-[16].

์••์ถ•์˜ ๊ณผ์ •์€ ํฌ๊ฒŒ ๋‘ ๊ฐ€์ง€๋กœ ์ด๋ฃจ์–ด์ง„๋‹ค. ์ฒซ ๋ฒˆ์งธ๋Š” ์ž…๋ ฅ๋œ ์˜์ƒ์— ๋Œ€ํ•œ ์ƒ๊ด€๋„๋ฅผ ์ œ๊ฑฐํ•˜๋Š” ์ „์ฒ˜๋ฆฌ ๊ณผ์ •์ด๋ฉฐ, ๋‘ ๋ฒˆ์งธ๋Š” ์‹ค์ œ ์••์ถ• ๊ณผ์ •์— ํ•ด๋‹นํ•˜๋Š” ์—”ํŠธ๋กœํ”ผ ์ฝ”๋”ฉ ๊ณผ์ •์ด๋‹ค. ๊ทธ์ค‘์—์„œ, ์ƒ๊ด€๋„ ์ œ๊ฑฐ ๊ณผ์ •์€ ๊ฐ ์˜์ƒ๋ผ๋ฆฌ ๊ฐ€์ง€๋Š” ์ƒ๊ด€๋„๋ฅผ ์ œ๊ฑฐํ•จ์œผ๋กœ์จ ์˜์ƒ ์••์ถ•์˜ ํšจ์œจ์„ ๋†’์ผ ์ˆ˜ ์žˆ๋‹ค. ๊ธฐ์กด์˜ 2์ฐจ์› ํ‰๋ฉด ์˜์ƒ๊ณผ๋Š” ๋‹ฌ๋ฆฌ ๋ฐ์ดํ„ฐ์˜ ์–‘์ด ๋Œ€ํญ ์ฆ๊ฐ€ํ•œ 3์ฐจ์› ์˜์ƒ์€ ๋ณ„๋„์˜ ์ „์ฒ˜๋ฆฌ ๊ณผ์ • ์—†์ด ์˜์ƒ์„ ๊ทธ๋Œ€๋กœ ์ „์†กํ•˜๊ธฐ์—๋Š” ๋น„์šฉ ๋ฉด์—์„œ ํฐ ์†ํ•ด์ด๋‹ค. ๊ทธ๋Ÿฌํ•œ ์ด์œ ๋กœ, ํšจ์œจ์ ์ธ ์••์ถ•์„ ์œ„ํ•ด ์ž…์ฒด ์˜์ƒ์— ๋Œ€ํ•œ ์ „์ฒ˜๋ฆฌ ๊ณผ์ •์ด ๋ถˆ๊ฐ€ํ”ผํ•˜๋‹ค. ํŠนํžˆ ๊ตญ์ œ ํ‘œ์ค€ํ™” ๊ธฐ๊ตฌ JPEG(Joint Photographic Experts Groups)์—์„œ๋Š” JPEG Pleno๋ผ๋Š” ์ด๋ฆ„์œผ๋กœ 3์ฐจ์› ์˜์ƒ ์••์ถ•์„ ์œ„ํ•œ ํ‘œ์ค€ํ™” ๊ณผ์ œ๋ฅผ ์ง„ํ–‰ ์ค‘์ด๋‹ค[17]-[18].

๋ณธ ๋…ผ๋ฌธ์—์„œ๋Š” EIA ์˜์ƒ์— ๋Œ€ํ•œ ์ „์ฒ˜๋ฆฌ ๊ณผ์ •์—์„œ EIA ์˜์ƒ์„ VIA ์˜์ƒ์œผ๋กœ ๋ณ€ํ™˜ํ•˜์—ฌ ์••์ถ•์˜ ํšจ์œจ์„ ์ฆ๋Œ€ํ•˜๋Š” ๋ฐ์— ์ฃผ๋ชฉํ•˜๊ณ  ์žˆ๋‹ค. ์ด๋Š” ๊ธฐ์กด์˜ ์ผ๋ฐ˜ ์˜์ƒ ์ „์ฒ˜๋ฆฌ ๊ณผ์ •๊ณผ๋Š” ๋‹ฌ๋ฆฌ EIA ์ด๋ฏธ์ง€๋ฅผ VIA ์ด๋ฏธ์ง€๋กœ ๋ณ€ํ™˜ํ•˜๋Š” ๊ณผ์ •์„ ์ถ”๊ฐ€ํ•จ์œผ๋กœ์จ ๊ธฐ์กด์˜ EIA๊ฐ€ ๊ฐ€์ง€๋Š” ๋ฌธ์ œ๋ฅผ ํ•ด๊ฒฐํ•˜์—ฌ ์••์ถ•๋ฅ ์„ ๋†’์ด๋Š” ๋ฐฉ๋ฒ•์ด๋‹ค. ๋˜ํ•œ, EIA์™€ VIA๋Š” ์„œ๋กœ ๋‹ค๋ฅธ ํŠน์„ฑ์„ ๊ฐ€์ง€๊ธฐ ๋•Œ๋ฌธ์— ๊ฐ ์˜์ƒ์— ์ ํ•ฉํ•œ ์••์ถ• ๋ฐฉ๋ฒ•์„ ์ฐพ๋Š” ์—ฐ๊ตฌ๊ฐ€ ํ•„์š”ํ•˜๋‹ค.

๋ณธ ๋…ผ๋ฌธ์€ ๋‹ค์Œ๊ณผ ๊ฐ™์ด ๊ตฌ์„ฑ๋˜์–ด ์žˆ๋‹ค. 1์ ˆ์—์„œ๋Š” ๋ณธ ๋…ผ๋ฌธ์—์„œ ๋‹ค๋ฃจ๋Š” 3์ฐจ์› ์ž…์ฒด ์˜์ƒ์— ๋Œ€ํ•œ ์†Œ๊ฐœ ๋ฐ ๋ณธ ๋…ผ๋ฌธ์˜ ํ•„์š”์„ฑ์„ ๋‹ค๋ฃจ์—ˆ๊ณ , 2์ ˆ์—์„œ๋Š” ๋ณธ ๋…ผ๋ฌธ์—์„œ ์‚ฌ์šฉ๋˜๋Š” EIA ๋ฐ VIA์— ๋Œ€ํ•œ ๊ธฐ๋ณธ ๊ฐœ๋… ๋ฐ ์‚ฌ์šฉํ•  ์••์ถ• ์ฝ”๋ฑ์— ๋Œ€ํ•œ ๋ถ„์„์„ ์ œ๊ณตํ•œ๋‹ค. 3์ ˆ์—์„œ๋Š” ๋ณธ ๋…ผ๋ฌธ์—์„œ ์ œ์•ˆํ•˜๋Š” ๋ฐฉ๋ฒ•์„ ๋‹ค๋ฃจ๋ฉฐ, 4์ ˆ์—์„œ๋Š” ํ•ด๋‹นํ•˜๋Š” ์‹คํ—˜ ๊ฒฐ๊ณผ ๋ฐ ๊ณ ์ฐฐ์— ๋Œ€ํ•ด ์–ธ๊ธ‰ํ•œ๋‹ค. ๋งˆ์ง€๋ง‰์œผ๋กœ 5์ ˆ์—์„œ๋Š” ๋ณธ ๋…ผ๋ฌธ์— ๋Œ€ํ•œ ๊ฒฐ๋ก ์œผ๋กœ ๋๋งบ์Œํ•œ๋‹ค.

2. ๊ธฐ๋ณธ ๊ฐœ๋…

๋ณธ ์—ฐ๊ตฌ์—์„œ ์‚ฌ์šฉํ•˜๊ณ ์ž ํ•˜๋Š” ์˜์ƒ์€ EIA(elemental image array) ์˜์ƒ์œผ๋กœ, ํ•˜๋‚˜์˜ ๋ Œ์ฆˆ๋ฅผ ํ†ตํ•ด ์ทจ๋“๋œ ์˜์ƒ์ด ์š”์†Œ ์˜์ƒ์ด๋‹ค. ์ด ์š”์†Œ ์˜์ƒ๋“ค์„ ๋ฐฐ์—ด๋กœ ๋‚˜์—ดํ•œ ๊ฒƒ์ด EIA์ด๋‹ค. EIA ์˜์ƒ์˜ ํฌ๊ธฐ๋Š” ๋ Œ์ฆˆ์˜ ํฌ๊ธฐ์™€ ๋ Œ์ฆˆ์˜ ๊ฐœ์ˆ˜์— ์˜ํ•ด ์ •ํ•ด์ง€๋ฉฐ, ๋ Œ์ฆˆ ๋ฐฐ์—ด์„ ํ†ตํ•ด ์–ป์–ด์ง€๋Š” ์˜์ƒ์ด๊ธฐ ๋•Œ๋ฌธ์— ์ผ๋ฐ˜ ์˜์ƒ๊ณผ๋Š” ๋‹ค๋ฅธ ํŠน์ง•์„ ๊ฐ€์ง„๋‹ค. ๊ทธ๋Ÿฌํ•œ ์ด์œ ๋กœ, EIA ์˜์ƒ์— ์ ํ•ฉํ•œ ์ƒ๊ด€๋„ ์ œ๊ฑฐ๋ฅผ ์œ„ํ•œ ๋ณ€ํ™˜ ๊ธฐ์ˆ ๊ณผ EIA ์˜์ƒ์„ ๋ณด๋‹ค ํšจ์œจ์ ์œผ๋กœ ์••์ถ•ํ•˜๊ธฐ ์œ„ํ•œ ์•Œ๊ณ ๋ฆฌ์ฆ˜์ด ์ ์šฉ๋˜์–ด์•ผ ํ•œ๋‹ค. ์ƒ๊ด€๋„ ์ œ๊ฑฐ๋ฅผ ์œ„ํ•œ ๋ณ€ํ™˜ ๊ธฐ์ˆ ์€ ์˜์ƒ์˜ ํฌ๊ธฐ์— ๋น„ํ•ด ๋ Œ์ฆˆ์˜ ํฌ๊ธฐ๊ฐ€ ์ž‘์„ ๋•Œ ์••์ถ• ํšจ์œจ์„ ๋†’์ด๊ธฐ ์œ„ํ•ด EIA ์˜์ƒ์„ VIA(view image array)๋กœ ๋ณ€ํ™˜ํ•˜๋Š” ๊ธฐ์ˆ ์„ ์ ์šฉํ•˜์˜€๋‹ค. ์ด์— ๋Œ€ํ•œ ์„ค๋ช…์€ ์•„๋ž˜์—์„œ ์ž์„ธํžˆ ๋‹ค๋ฃจ๋„๋ก ํ•œ๋‹ค. ์••์ถ• ์ฝ”๋ฑ์œผ๋กœ ์‚ฌ์šฉํ•œ ์•Œ๊ณ ๋ฆฌ์ฆ˜์€ ๊ธฐ์กด์˜ H.263, MPEG-4๊ฐ€ ๊ฐ€์ง€๋Š” ํ•œ๊ณ„์„ฑ์„ ํ•ด๊ฒฐํ•˜์—ฌ ๋ชจ๋ฐ”์ผ ๋™์˜์ƒ ์„œ๋น„์Šค ๋“ฑ ํ˜„์žฌ ๋ฉ€ํ‹ฐ๋ฏธ๋””์–ด ์ „์†ก ๋ถ„์•ผ์—์„œ ์ฃผ๋กœ ์‚ฌ์šฉ๋˜๋Š” H.264 ์ฝ”๋ฑ์„ ์‚ฌ์šฉํ•˜์—ฌ ์˜์ƒ์„ ์••์ถ•ํ•˜์˜€๋‹ค. H.264์— ๋Œ€ํ•œ ๋ถ„์„์€ ๋‹ค์Œ ์ ˆ์—์„œ ์ด์–ด์ง„๋‹ค.

2.1 EIA / VIA

์•ž์„œ ์–ธ๊ธ‰ํ•œ ๋ฐ”์™€ ๊ฐ™์ด, EIA ์˜์ƒ์€ ๋‹ค์ˆ˜์˜ ๋ Œ์ฆˆ ๋ฐฐ์—ด์„ ํ†ตํ•ด ์–ป์–ด์ง„ ์˜์ƒ์œผ๋กœ, 2์ฐจ์› ์š”์†Œ ์˜์ƒ๋“ค์˜ 2์ฐจ์› ๋ฐฐ์—ด ํ˜•ํƒœ๋กœ ์ด๋ฃจ์–ด์ง„ ์˜์ƒ์ด๋‹ค. ๋ Œ์ฆˆ ๋ฐฐ์—ด์„ ํ†ตํ•ด ํš๋“๋œ ์˜์ƒ์œผ๋กœ, ๊ฐ ์˜์ƒ์€ 3์ฐจ์› ๋ฌผ์ฒด์— ๋Œ€ํ•œ ํŠน์ • ์‹œ์ ์— ๋Œ€ํ•œ ์ •๋ณด๋ฅผ ๊ฐ€์ง€๊ณ  ์žˆ๋‹ค๋Š” ์ ์—์„œ ๊ธฐ์กด ์˜์ƒ๊ณผ ๋‹ค๋ฅธ ํŠน์ง•์„ ๊ฐ€์ง€๊ณ  ์žˆ๋‹ค. ์˜์ƒ์—์„œ ๊ฐ€๊นŒ์ด ์žˆ๋Š” ๋ Œ์ฆˆ์— ์˜ํ•ด ์ทจ๋“๋œ ์˜์ƒ์ผ์ˆ˜๋ก 3์ฐจ์› ๋ฌผ์ฒด์— ๋Œ€ํ•œ ์ •๋ณด๊ฐ€ ์ง‘์ค‘์ ์œผ๋กœ ๋ชจ์—ฌ ์žˆ๊ณ  ๋ฉ€๋ฆฌ ์žˆ๋Š” ๋ Œ์ฆˆ์— ์˜ํ•ด ์ทจ๋“๋ ์ˆ˜๋ก ์ •๋ณด๊ฐ€ ๋ถ„์‚ฐ๋˜์–ด์žˆ๋Š” ํ˜•ํƒœ์ด๋‹ค.

์š”์†Œ ์˜์ƒ์˜ ํฌ๊ธฐ์™€ ๋ Œ์ฆˆ์˜ ํฌ๊ธฐ์— ์˜ํ•ด ์ทจ๋“๋˜๋Š” ์˜์ƒ์˜ ํฌ๊ธฐ๊ฐ€ ๋ณ€๋™๋˜๊ธฐ ๋•Œ๋ฌธ์— ๊ฐ ์š”์†Œ ์˜์ƒ์˜ ํฌ๊ธฐ๊ฐ€ ๋งค์šฐ ์ž‘๊ณ  ๋ Œ์ฆˆ์˜ ๊ฐœ์ˆ˜๊ฐ€ ์˜์ƒ ํฌ๊ธฐ์— ๋น„ํ•ด ๋งค์šฐ ํฌ๋‹ค๋ฉด, ์ด ์˜์ƒ์€ ์••์ถ•์— ์–ด๋ ค์›€์„ ๊ฒช์„ ์ˆ˜ ์žˆ๋‹ค. ์™œ๋ƒํ•˜๋ฉด, ๊ฐ ์š”์†Œ ์˜์ƒ์ด ๋งค์šฐ ์ž‘๊ธฐ ๋•Œ๋ฌธ์— ์••์ถ•์˜ ํ•œ๊ณ„์— ๊ธˆ๋ฐฉ ๋„๋‹ฌํ•˜๊ฒŒ ๋˜๋ฉฐ ์˜์ƒ๋งˆ๋‹ค ๊ฐ€์ง€๊ฒŒ ๋˜๋Š” ์˜ค๋ฒ„ํ—ค๋“œ์˜ ํฌ๊ธฐ๊ฐ€ ์••์ถ• ๋ฐ์ดํ„ฐ์— ๋น„ํ•ด ํ„ฑ์—†์ด ํฌ๋‹ค. ์ด๋Ÿฌํ•œ ์ ์€ ์••์ถ•๋ฅ ์„ ์ €๊ฐ์‹œํ‚ค๋Š” ์š”์ธ์ด ๋œ๋‹ค. ์ด๋Ÿฌํ•œ ์ด์œ ๋กœ, ์š”์†Œ ์˜์ƒ์˜ ํฌ๊ธฐ๊ฐ€ ์ž‘์€ EIA์˜ ๊ฒฝ์šฐ, EIA๋ฅผ VIA ์˜์ƒ์œผ๋กœ ๋ณ€ํ™˜ํ•˜์—ฌ ์š”์†Œ ์˜์ƒ์˜ ํฌ๊ธฐ๋ฅผ ํ‚ค์šธ ์ˆ˜ ์žˆ๋‹ค.

VIA๋Š” ๊ฐ ์š”์†Œ ์˜์ƒ์—์„œ ๊ฐ™์€ ์‹œ์ ์˜ ํ”ฝ์…€๋“ค์„ ๋ชจ์•„ ํ•˜๋‚˜์˜ ์š”์†Œ ์˜์ƒ์œผ๋กœ ์ƒ์„ฑํ•œ ๊ฒƒ์ด๋‹ค[19]. EIA๋ฅผ VIA๋กœ ๋ณ€ํ™˜ํ•˜๋Š” ์‹์€ ๋‹ค์Œ๊ณผ ๊ฐ™๋‹ค.

(1)
$$EIA(i,\:j,\:r,\:c)\Rightarrow VIA(r,\:c,\:i,\:j)$$

์—ฌ๊ธฐ์„œ, ์‹ (1)์„ ํ†ตํ•ด EIA์˜ ๊ฐ ์š”์†Œ ์˜์ƒ์—์„œ ๊ฐ™์€ ์‹œ์ ์˜ ํ”ฝ์…€์„ ๋ชจ์•„ ํ•˜๋‚˜์˜ ์š”์†Œ ์˜์ƒ์œผ๋กœ ์ƒ์„ฑํ•  ์ˆ˜ ์žˆ๋‹ค. ๋ณ€ํ™˜๋œ VIA๋Š” ๊ธฐ์กด์˜ EIA์— ๋น„ํ•ด ๋ฌผ์ฒด์˜ ๊นŠ์ด์ •๋ณด์™€ ์ƒ๊ด€์—†์ด ๊ฐ ์š”์†Œ ์˜์ƒ์— ๋ณด์—ฌ์ง€๋Š” ๋ฌผ์ฒด์˜ ๋ชจ์–‘์ด ์„œ๋กœ ๋†’์€ ์œ ์‚ฌ์„ฑ์„ ๋ณด์ธ๋‹ค. EIA๋ฅผ ํ†ตํ•ด VIA๋ฅผ ์ƒ์„ฑํ•˜๋Š” ๋ฐฉ๋ฒ•์€ ์•„๋ž˜ ๊ทธ๋ฆผ 1๊ณผ ๊ฐ™๋‹ค.

๊ทธ๋ฆผ. 1. EIA ์˜์ƒ์— ๋Œ€ํ•œ VIA ๋ณ€ํ™˜ ๊ณผ์ •

Fig. 1. EIA to VIA transform

../../Resources/kiee/KIEE.2020.69.1.197/fig1.png

์š”์†Œ ์˜์ƒ์˜ ํฌ๊ธฐ๋Š” ์ž‘๊ณ  ๋ Œ์ฆˆ์˜ ๊ฐœ์ˆ˜๊ฐ€ ๋งŽ์€ EIA๋ฅผ VIA๋กœ ๋ณ€ํ™˜ํ•˜๊ฒŒ ๋˜๋ฉด ์š”์†Œ ์˜์ƒ์˜ ํฌ๊ธฐ๋Š” ์ปค์ง€๋Š” ๋ฐ˜๋ฉด, ๋ Œ์ฆˆ์˜ ๊ฐœ์ˆ˜๊ฐ€ ์ค„์–ด๋“ค๊ฒŒ ๋œ๋‹ค. ๊ทธ๋ ‡๊ฒŒ ๋˜๋ฉด, ๊ธฐ์กด์˜ EIA๋ณด๋‹ค ์š”์†Œ ์˜์ƒ์˜ ํฌ๊ธฐ๊ฐ€ ์ปค์ง€๊ณ , ๊ฐ ์š”์†Œ ์˜์ƒ ๊ฐ„ ์ƒ๊ด€๋„๊ฐ€ ๋†’์•„์ง€๋ฏ€๋กœ ์˜ˆ์ธก ๊ณผ์ •์—์„œ ์‰ฝ๊ฒŒ ์ƒ๊ด€๋„ ์ œ๊ฑฐ๊ฐ€ ๊ฐ€๋Šฅํ•˜๋‹ค. ์œ„์™€ ๊ฐ™์€ ๊ณผ์ •์„ ํ†ตํ•ด ์š”์†Œ ์˜์ƒ์˜ ํฌ๊ธฐ๊ฐ€ ์ž‘์€ EIA ์˜์ƒ์„ VIA ์˜์ƒ์œผ๋กœ ๋ณ€ํ™˜ํ•จ์œผ๋กœ์จ ๋ณ€ํ™˜๋œ ์˜์ƒ์€ ์••์ถ•์— ๋ณด๋‹ค ์ ํ•ฉํ•œ ํ˜•ํƒœ๋ฅผ ๊ฐ–์ถ”๊ฒŒ ๋œ๋‹ค. ํ•˜์ง€๋งŒ ์š”์†Œ ์˜์ƒ์˜ ํฌ๊ธฐ์™€ ๋ Œ์ฆˆ์˜ ๊ฐœ์ˆ˜๊ฐ€ ๋™์ผํ•œ ๊ฒฝ์šฐ, ์˜์ƒ์„ EIA ํ˜น์€ VIA๋กœ ๋ณ€ํ™˜ํ•˜๋Š” ๊ฒƒ์ด ์••์ถ•๋ฅ ์„ ๋†’์ธ๋‹ค๊ณ  ํ•  ์ˆ˜ ์—†๋‹ค. ๊ทธ๋Ÿฌํ•œ ์ด์œ ๋กœ, ์˜์ƒ์˜ ํŠน์„ฑ์„ ๋ถ„์„ํ•œ ํ›„, ๊ฐ ์˜์ƒ์— ์ ํ•ฉํ•œ ๋ณ€ํ™˜ ๊ณผ์ •์„ ๊ฑฐ์นœ ํ›„ ์••์ถ•์— ์ ์šฉํ•˜๋Š” ๊ณผ์ •์ด ํ•„์š”ํ•˜๋‹ค.

2.2 H.264/AVC ์••์ถ• ์ฝ”๋ฑ

ํ˜„์žฌ ์ฃผ๋ชฉ๋ฐ›๊ณ  ์žˆ๋Š” H.264/AVC ์ฝ”๋ฑ์€ ํ˜„์žฌ ๊ณ ์„ ๋ช… ๋น„๋””์˜ค์— ๋Œ€ํ•œ ๊ณ ํšจ์œจ์˜ ์••์ถ• ํฌ๋งท์„ ์ œ๊ณตํ•œ๋‹ค[20]-[23]. ๋ธ”๋ฃจ๋ ˆ์ด ๋””์Šคํฌ๋ฅผ ํฌํ•จํ•œ ์ธํ„ฐ๋„ท ์ŠคํŠธ๋ฆฌ๋ฐ ์„œ๋น„์Šค๋ฅผ ์ œ๊ณตํ•˜๋Š” ์—ฌ๋Ÿฌ ์†Œํ”„ํŠธ์›จ์–ด์—์„œ ์‚ฌ์šฉํ•˜๊ณ  ์žˆ๋Š” H.264๋Š” ๊ธฐ์กด์˜ MPEG-4์— ๋น„ํ•ด ์›”๋“ฑํžˆ ์ฆ๊ฐ€ํ•œ ์ „์†ก ์†๋„์™€ ๋†’์€ ์••์ถ•๋ฅ ๋กœ ํ˜„์žฌ ๋Œ€๋ถ€๋ถ„์˜ ์ŠคํŠธ๋ฆฌ๋ฐ ์„œ๋น„์Šค ๋ฐ ์ „์†ก ์„œ๋น„์Šค์—์„œ ์‚ฌ์šฉ๋˜๊ณ  ์žˆ๋‹ค. ํŠนํžˆ, ๊ธฐ์กด์˜ ์••์ถ• ์ฝ”๋ฑ์— ๋น„ํ•ด ์ธ์ฝ”๋”ฉ ๋ฐ ๋””์ฝ”๋”ฉ ๊ณผ์ •์—์„œ ๋ณต์žก๋„๊ฐ€ ๋†’๊ธฐ ๋•Œ๋ฌธ์— ๋†’์€ ์••์ถ•๋ฅ ์„ ์ž๋ž‘ํ•  ์ˆ˜ ์žˆ๋‹ค. ๊ทธ ์ค‘, ๋ณธ ์—ฐ๊ตฌ์—์„œ๋Š” H.264/AVC๊ฐ€ ์ œ๊ณตํ•˜๋Š” ์••์ถ• ํ”„๋กœํŒŒ์ผ ์ค‘, baseline profile ๋ฐ high profile์— ๋Œ€ํ•ด ๋ถ„์„ํ•˜์˜€๋‹ค[20].

๋จผ์ €, baseline profile์˜ ๊ฒฝ์šฐ ์ด๋™ ํ†ต์‹ , ํ™”์ƒํšŒ์˜ ๋“ฑ ์ €๋Œ€์—ญํญ์—์„œ ์ฃผ๋กœ ํ™œ์šฉํ•˜๋Š” ํ”„๋กœํŒŒ์ผ๋กœ, H.264์˜ ๊ฐ€์žฅ ๊ธฐ๋ณธ์ ์ธ ํ”„๋กœํŒŒ์ผ์ด๋ผ ํ•  ์ˆ˜ ์žˆ๋‹ค. ์ „์†ก ๋ฐ์ดํ„ฐ์— ๋Œ€ํ•œ ์—๋Ÿฌ ๊ฐ•์ธ์„ฑ์— ๋Œ€ํ•œ ๋ถ€๋‹ด์ด ์ ์œผ๋ฉฐ, ๊ณ ํ’ˆ์งˆ์— ๋Œ€ํ•œ ๋ฏผ๊ฐ์„ฑ์ด ๋–จ์–ด์ง€๋Š” ๋ถ„์•ผ์—์„œ ์‚ฌ์šฉ๋˜๋Š” ๋งŒํผ, baseline profile์ด ๊ฐ€์ง€๋Š” ์••์ถ• ์˜ต์…˜์— ๋Œ€ํ•œ ์ •๋ณด๋Š” ๋งค์šฐ ๋‹จ์ˆœํ•˜๋ฉฐ ์••์ถ•์˜ ์ •๋„๊ฐ€ ๋†’๋‹ค. ์ด๋Ÿฌํ•œ ์ด์œ ๋กœ baseline profile์€ ์˜์ƒ ํ™”์งˆ์— ๋Œ€ํ•œ ๋ฏผ๊ฐ๋„๊ฐ€ ๋–จ์–ด์ง€๋Š” ๋ถ„์•ผ์— ์ฃผ๋กœ ์‚ฌ์šฉ๋œ๋‹ค.

๊ทธ์™€ ๋‹ฌ๋ฆฌ, high profile์€ ๊ณ ํ’ˆ์งˆ ๋ฐ ๊ณ ์„ฑ๋Šฅ์— ์ดˆ์ ์„ ๋‘์–ด HD๊ธ‰ TV ๋ฐฉ์†ก, ์ŠคํŠธ๋ฆฌ๋ฐ ๋น„๋””์˜ค ์„œ๋น„์Šค ๋“ฑ ๋†’์€ ์ˆ˜์š”๋งŒ์กฑ๋„๋ฅผ ์š”๊ตฌํ•˜๋Š” ๋ถ„์•ผ์— ์ ํ•ฉํ•œ profile์ด๋‹ค. ์ดˆ๊ธฐ ๋””์ž์ธ์œผ๋กœ ์ €ํ•ด์ƒ๋„ ๋ฐ ๋ชจ๋ฐ”์ผ์šฉ์— ์ ํ•ฉํ•œ baseline profile๊ณผ๋Š” ๋‹ค๋ฅด๊ฒŒ ๊ณ ํ•ด์ƒ๋„์˜ ๋ฐฉ์†ก์šฉ ์˜์ƒ๊ณผ ์ €์žฅ์šฉ์œผ๋กœ ์‚ฌ์šฉ๋˜๊ธฐ ๋•Œ๋ฌธ์— baseline profile๋ณด๋‹ค ๋ฐ์ดํ„ฐ์— ๋Œ€ํ•œ ์—๋Ÿฌ ๊ฐ•์ธ์„ฑ์ด ๊ฐ•ํ•˜๋ฉฐ ์••์ถ•์— ์‚ฌ์šฉ๋˜๋Š” ์˜ต์…˜ ์ •๋ณด ์—ญ์‹œ ๋ฌด์ˆ˜ํžˆ ๋งŽ๋‹ค. ๊ทธ ์˜ˆ๋กœ, RGB value 128๋กœ๋งŒ ์ด๋ฃจ์–ด์ง„ ๋‹จ์ƒ‰ ์˜์ƒ์„ baseline profile ๋ฐ high profile์— ์ ์šฉํ•˜์—ฌ ์••์ถ•ํ•œ ๊ฒฐ๊ณผ, high profile ์˜ค๋ฒ„ํ—ค๋“œ ์ •๋ณด์˜ ํฌ๊ธฐ๊ฐ€ baseline profile์— ๋Œ€ํ•œ ํฌ๊ธฐ์— ๋น„ํ•ด ์•ฝ 3๋ฐฐ ํฌ๋‹ค๋Š” ์ ์„ ํ™•์ธํ•  ์ˆ˜ ์žˆ์—ˆ๋‹ค. ์ด๋Š” ํ”„๋กœํŒŒ์ผ์ด ๋†’์•„์งˆ์ˆ˜๋ก, ์˜์ƒ์„ ์••์ถ•ํ•˜๋Š”๋ฐ ์‚ฌ์šฉ๋˜๋Š” ์˜ต์…˜ ์ •๋ณด๊ฐ€ ๋งŽ์•„์ง์œผ๋กœ์จ ์˜์ƒ๋งˆ๋‹ค ๊ฐ€์ง€๋Š” ์˜ค๋ฒ„ํ—ค๋“œ ์ •๋ณด๊ฐ€ ์ฆ๊ฐ€ํ•œ๋‹ค๋Š” ๊ฒƒ์„ ์˜๋ฏธํ•œ๋‹ค. ํŠนํžˆ, ์˜์ƒ์˜ ํฌ๊ธฐ๊ฐ€ ์ž‘์•„์งˆ์ˆ˜๋ก, ์••์ถ•๋œ ๋ฐ์ดํ„ฐ๋ณด๋‹ค ๊ฐ ์š”์†Œ ์˜์ƒ์ด ๊ฐ€์ง€๋Š” ์˜ค๋ฒ„ํ—ค๋“œ ์ •๋ณด๊ฐ€ ๋” ๋งŽ์•„์ ธ ์••์ถ• ์„ฑ๋Šฅ์„ ์ค„์ด๋Š” ์•…์˜ํ–ฅ์„ ๋ผ์น  ์ˆ˜ ์žˆ์Œ์„ ์•”์‹œํ•œ๋‹ค.

3. ์ œ์•ˆํ•˜๋Š” ๋ฐฉ๋ฒ•

๋ณธ ์—ฐ๊ตฌ์—์„œ๋Š” ์š”์†Œ ์˜์ƒ ๋ฐฐ์—ด์˜ ํšจ์œจ์ ์ธ ์••์ถ•์„ ์œ„ํ•˜์—ฌ EIA ์˜์ƒ์„ VIA ์˜์ƒ์œผ๋กœ ๋ณ€ํ™˜ํ•˜๋Š” ์˜ต์…˜ ๊ณผ์ •์„ ๊ฑฐ์นœ ํ›„, 1์ฐจ์› ์‹œํ€€์Šค ๋ฐฐ์—ด๋กœ ์ €์žฅํ•˜์—ฌ H.264/AVC ์ฝ”๋ฑ์— ์ ์šฉํ•˜๋Š” ๊ณผ์ •์œผ๋กœ ์ง„ํ–‰ํ•œ๋‹ค. 1์ฐจ์› ์‹œํ€€์Šค ๋ฐฐ์—ด๋กœ ์ €์žฅ๋œ ๋ฐ์ดํ„ฐ๋Š” YUV 4:2:0 ํฌ๋งท์œผ๋กœ ์ด๋ฃจ์–ด์กŒ๋‹ค. ์ด๋•Œ, ์ž…๋ ฅ ์˜์ƒ์— ๋Œ€ํ•œ ๋ณ€ํ™˜ ๊ธฐ๋ฒ•๊ณผ ์‚ฌ์šฉํ•˜๋Š” ์••์ถ• ์ฝ”๋ฑ์˜ ํ”„๋กœํŒŒ์ผ์„ ๋‹ค๋ฅด๊ฒŒ ์ ์šฉํ•˜์—ฌ ์˜์ƒ์— ์ ํ•ฉํ•œ ์••์ถ• ๊ณผ์ •์„ ์ฐพ๊ณ ์ž ํ•œ๋‹ค. ๋ณธ ์—ฐ๊ตฌ์—์„œ ์ง„ํ–‰ํ•œ ์‹คํ—˜ ๊ณผ์ •์€ ๋‹ค์Œ ๊ทธ๋ฆผ 2์™€ ๊ฐ™๋‹ค.

H.264 ์ฝ”๋ฑ์— ์š”์†Œ ์˜์ƒ์„ ์ ์šฉํ•˜๊ธฐ ์œ„ํ•ด์„œ๋Š”, 2์ฐจ์›์œผ๋กœ ์ด๋ฃจ์–ด์ง„ ๋ฐ์ดํ„ฐ๋ฅผ 1์ฐจ์› ์‹œํ€€์Šค ๋ฐ์ดํ„ฐ๋กœ ๋‚˜์—ดํ•ด์•ผ ํ•œ๋‹ค. EIA ์˜์ƒ์„ ๋‹จ์ผํ•œ ์ •์ง€ ์˜์ƒ์œผ๋กœ ๊ฐ„์ฃผํ•˜๊ณ  ์••์ถ•์„ ์ง„ํ–‰ํ•  ์ˆ˜๋„ ์žˆ์ง€๋งŒ, ์ด ๋ฐฉ๋ฒ•์€ ์š”์†Œ ์˜์ƒ ๊ฐ„์˜ ๋ถˆ์—ฐ์†์„ฑ ๋•Œ๋ฌธ์— ์••์ถ•์˜ ํšจ์œจ์ด ๋–จ์–ด์ง„๋‹ค. ์ด๋Ÿฌํ•œ ์ด์œ ๋กœ 2์ฐจ์› ๋ฐฐ์—ด๋กœ ์ด๋ฃจ์–ด์ง„ EIA ์˜์ƒ์„ 1์ฐจ์› ์Šค์บ” ๋ฐฉ๋ฒ•์„ ์ด์šฉํ•˜์—ฌ ์‹œํ€€์Šค ํ˜•ํƒœ๋กœ ๋ณ€ํ˜•ํ•˜์—ฌ ์••์ถ•์„ ์ง„ํ–‰ํ•œ๋‹ค. ์ด ๋ฐฉ๋ฒ•์€ S. Yeom et al์— ์˜ํ•ด ๊ณ ์•ˆ๋œ ๋ฐฉ๋ฒ•์œผ๋กœ ์ง‘์  ์˜์ƒ์˜ ํŠน์ง•์„ ๊ณ ๋ คํ•œ ๋ฐฉ๋ฒ•์ด๋‹ค[24]. ๊ทธ ์ค‘ ๋ณธ ๋…ผ๋ฌธ์—์„œ๋Š” ๋‚˜์„ ํ˜• ์Šค์บ” ๋ฐฉ์‹์„ ์„ ํƒํ•˜์˜€๋‹ค. ์ž์„ธํ•œ ์Šค์บ” ๋ฐฉ์‹์€ ๋‹ค์Œ ๊ทธ๋ฆผ 3๊ณผ ๊ฐ™๋‹ค.

๋‚˜์„  ์Šค์บ” ๋ฐฉ๋ฒ•์€ ์ง‘์  ์˜์ƒ์˜ ๊ฒฝ๊ณ„์— ์œ„์น˜ํ•œ ์š”์†Œ ์˜์ƒ์ด ์—ฐ์†์ ์ธ ์˜์ƒ์œผ๋กœ ์ž…๋ ฅ๋˜๋Š” ๊ฒƒ์„ ๋ฐฉ์ง€ํ•˜๊ณ , ๋ถ€๋ถ„์ ์œผ๋กœ ์ˆ˜์ง๋ฐฉํ–ฅ์˜ ์ƒ๊ด€๋„๋ฅผ ์ œ๊ฑฐํ•  ์ˆ˜ ์žˆ๋‹ค๋Š” ์ ์—์„œ ์••์ถ• ํšจ์œจ์„ ๋†’์ผ ์ˆ˜์žˆ๋‹ค. ๊ทธ๋Ÿฌ๋‚˜ ๋‚˜์„ ํ˜• ์Šค์บ” ๋ฐฉ๋ฒ•์€ 1์ฐจ์› ์Šค์บ” ๋ฐฉ๋ฒ•์ด๋ฏ€๋กœ ์ˆ˜์ง ๋ฐ ์ˆ˜ํ‰ ๋ฐฉํ–ฅ์˜ ์ƒ๊ด€๋„๋ฅผ ๋™์‹œ์— ๊ณ ๋ คํ•˜์ง€ ๋ชปํ•œ๋‹ค. ๋˜ํ•œ, ์Šค์บ” ์ˆœ์„œ์— ์˜ํ•ด ๋ถ€๋ถ„์ ์œผ๋กœ ์ˆ˜์ง ๋˜๋Š” ์ˆ˜ํ‰ ๋ฐฉํ–ฅ์˜ ์ƒ๊ด€๋„๋ฅผ ๊ณ ๋ คํ•˜๊ธฐ ๋•Œ๋ฌธ์— ๋ถ€ํ˜ธํ™” ํšจ์œจ์„ ํ–ฅ์ƒ์‹œํ‚ค๋Š”๋ฐ ํ•œ๊ณ„๊ฐ€ ์žˆ๋‹ค.

๊ทธ๋ฆผ. 2. ์ œ์•ˆํ•˜๋Š” ๋ฐฉ๋ฒ•์— ๋Œ€ํ•œ ํ๋ฆ„๋„

Fig. 2. Flowchart for proposed method

../../Resources/kiee/KIEE.2020.69.1.197/fig2.png

๊ทธ๋ฆผ. 3. ์ œ์•ˆ๋œ ๋ฐฉ๋ฒ•์—์„œ ์‚ฌ์šฉ๋œ ๋‚˜์„ ํ˜• ์Šค์บ” ๊ธฐ๋ฒ•

Fig. 3. MPEG-4 spiral scanning method

../../Resources/kiee/KIEE.2020.69.1.197/fig3.png

๋‚˜์„  ์Šค์บ”์ด ์ด๋ฃจ์–ด์ง„ 1์ฐจ์›์œผ๋กœ ๋œ ์‹œํ€€์Šค ๋ฐ์ดํ„ฐ๋Š” ๋ณธ ๋…ผ๋ฌธ์—์„œ ์‚ฌ์šฉํ•˜๊ณ ์ž ํ•˜๋Š” H.264/AVC ์˜ ์ž…๋ ฅ ํฌ๋งท์— ๋งž์ถ”๊ธฐ ์œ„ํ•ด YUV ํ˜•ํƒœ๋กœ ๋ณ€ํ™˜๋˜์–ด์•ผ ํ•œ๋‹ค. ์ด๋•Œ ์‹ (2)๋ฅผ ์ ์šฉํ•˜์—ฌ ๋ณ€ํ™˜ํ•˜์˜€๋‹ค.

(2)
$$ \left[\begin{array}{l} {y} \\ {u} \\ {v} \end{array}\right]=\left[\begin{array}{ccc} {0.299} & {0.587} & {0.114} \\ {-0.148} & {-0.289} & {0.437} \\ {0.615} & {-0.515} & {-0.00} \end{array}\right]\left[\begin{array}{l} {r} \\ {g} \\ {b} \end{array}\right] $$

์œ„ ์‹์„ ํ†ตํ•ด ๋ณ€ํ™˜๋œ YUV ์˜์ƒ์€ U์™€ V ์„ฑ๋ถ„์„ ๊ฐ€๋กœ, ์„ธ๋กœ ๋ฐฉํ–ฅ์œผ๋กœ 2:1 down sampling ํ•˜์—ฌ 4:2:0 ํ˜•ํƒœ๋กœ ์ฒ˜๋ฆฌํ•œ๋‹ค. U์™€ V ์„ฑ๋ถ„์˜ ์œ„์น˜๋Š” MPEG-2์˜ ๋ฐฉ์‹์„ ๋”ฐ๋ผ ์„ ํƒํ•˜์˜€๋‹ค[25]. ์••์ถ• ์„ฑ๋Šฅ์„ ํ‰๊ฐ€ํ•˜๊ธฐ ์œ„ํ•ด์„œ ์›๋ณธ๋ฐ์ดํ„ฐ์™€ ์••์ถ•๋œ ํŒŒ์ผ์˜ ํฌ๊ธฐ๋ฅผ ๋น„๊ตํ•˜์—ฌ ์••์ถ•๋ฅ ์„ ๊ณ„์‚ฐํ•˜์˜€๋‹ค. ์••์ถ•๋ฅ ์— ๊ด€ํ•œ ์‹์€ ๋‹ค์Œ ์‹ (3)๊ณผ ๊ฐ™๋‹ค.

(3)
$\begin{align*} r=\dfrac{Original \enspace file \enspace size(bytes)}{Compressed \enspace file \enspace size(bytes)}\\ \end{align*}$

๋‹ค์Œ์œผ๋กœ, ์••์ถ•๋œ ํŒŒ์ผ์„ ๋ณต์›ํ•œ ํ›„, ๋‹ค์‹œ 2์ฐจ์› ์š”์†Œ ์˜์ƒ ๋ฐฐ์—ด์˜ ํ˜•ํƒœ๋กœ ์žฌ๋ฐฐ์—ดํ•˜์—ฌ color PSNR(Peak signal-to-noise ratio)์„ ๊ณ„์‚ฐํ•˜์—ฌ ์„ฑ๋Šฅ์„ ํ‰๊ฐ€ํ•˜์˜€๋‹ค. ์‹ (4)๋Š” color PSNR์„ ๊ตฌํ•˜๋Š” ์‹์ด๋ฉฐ, color PSNR์€ R, G, B ๊ฐ๊ฐ์˜ ์˜์ƒ์— ๋Œ€ํ•œ PSNR ๊ฐ’์˜ ํ‰๊ท ๊ฐ’์ด๋‹ค. ์ด๋•Œ, ์‹ (4)๋Š” ์š”์†Œ ์˜์ƒ์˜ R, G, B ๊ฐ๊ฐ์˜ MSE(Mean Squared Error) ๊ฐ’์— ์˜ํ•ด ์ •์˜๋œ๋‹ค. ์‹ (5)๋Š” R,G,B ๊ฐ๊ฐ์— ๋Œ€ํ•œ MSE ๊ฐ’๊ณผ color PSNR์— ํ•„์š”ํ•œ MSE๋ฅผ ๊ตฌํ•˜๋Š” ์‹์ด๋‹ค. ์ด๋•Œ, $O$๋Š” $M\times N$ ํฌ๊ธฐ์˜ ์›๋ณธ ์˜์ƒ์„ ๋‚˜ํƒ€๋‚ด๊ณ , $R$์€ ๋ณต์›๋œ ์˜์ƒ์„ ๋‚˜ํƒ€๋‚ธ๋‹ค.

(4)
$$CPSNR=10\log 10(\dfrac{255^{2}}{MSE})$$

(5)
$$ \begin{aligned} &M S E_{R}=\frac{1}{M N} \sum_{i=0}^{M-1} \sum_{j=0}^{N-1}\left|O_{R}(i, j)-R_{R}(i, j)\right|^{2}\\ &M S E_{G}=\frac{1}{M N} \sum_{i=0}^{M-1} \sum_{j=0}^{N-1}\left|O_{G}(i, j)-R_{G}(i, j)\right|^{2}\\ &M S E_{B}=\frac{1}{M N} \sum_{i=0}^{M-1} \sum_{j=0}^{N-1}\left|O_{B}(i, j)-R_{B}(i, j)\right|^{2}\\ &M S E=\frac{M S E_{R}+M S E_{G}+M S E_{B}}{3} \end{aligned} $$

4. ์‹คํ—˜ ๊ฒฐ๊ณผ

์‹คํ—˜์—์„œ ์‚ฌ์šฉ๋œ ๋ฐ์ดํ„ฐ๋Š” ๋‹ค์Œ๊ณผ ๊ฐ™๋‹ค. ๊ทธ๋ฆผ 4๋Š” ์‹ค์ œ EIA ์˜์ƒ์„ VIA ์˜์ƒ์œผ๋กœ ๋ณ€ํ™˜ํ•œ ์‹คํ—˜์ด๋‹ค. ๊ทธ๋ฆผ 4์—์„œ ๋ณด์ด๋Š” ๋ฐ”์™€ ๊ฐ™์ด, (a)๋Š” EIA ๋ฐ์ดํ„ฐ์ด๋ฉฐ, (b)๋Š” EIA ๋ฐ์ดํ„ฐ๋ฅผ VIA ๋ฐ์ดํ„ฐ๋กœ ๋ณ€ํ™˜ํ•œ ์˜์ƒ์ด๋‹ค. EIA์˜ ๊ฒฝ์šฐ ๊ฐ ๋ Œ์ฆˆ์— ๋Œ€ํ•œ ์‹œ์  ์ •๋ณด๊ฐ€ ๋‹ด๊ฒจ ์žˆ๊ธฐ ๋•Œ๋ฌธ์—, ๋ Œ์ฆˆ๊ฐ€ ๋ฐ”๋ผ๋ณด๋Š” ๊ฐ๋„๋งˆ๋‹ค ์˜์ƒ์ด ๋‹ค๋ฅด๊ฒŒ ๋ณด์ธ๋‹ค๋Š” ์ ์„ ํ™•์ธํ•  ์ˆ˜ ์žˆ๋‹ค. ๊ทธ์— ๋ฐ˜ํ•ด VIA ์˜์ƒ์˜ ๊ฒฝ์šฐ, ๋ชจ๋“  ๋ Œ์ฆˆ์˜ ๋™์ผํ•œ ์‹œ์ ์— ๋Œ€ํ•œ ํ”ฝ์…€ ์ •๋ณด๋ฅผ ๋ชจ์•„ ์ƒˆ๋กœ์šด ์˜์ƒ์œผ๋กœ ๋ณ€ํ™˜ํ•˜์˜€๊ธฐ ๋•Œ๋ฌธ์— ๋Œ€์ฒด๋กœ ์œ ์‚ฌํ•œ ์˜์ƒ์œผ๋กœ ๋ณด์ž„์„ ์•Œ ์ˆ˜ ์žˆ๋‹ค. ๊ฐ ์˜์ƒ์— ๋Œ€ํ•œ ์ •๋ณด๋Š” ๋‹ค์Œ ํ‘œ 1๊ณผ ๊ฐ™๋‹ค. ํ‘œ 1์—์„œ ๋ณด์ด๋Š” ๋ฐ”์™€ ๊ฐ™์ด ์š”์†Œ ์˜์ƒ์˜ ํฌ๊ธฐ์™€ ์š”์†Œ ์˜์ƒ์˜ ๊ฐœ์ˆ˜๊ฐ€ ๋™์ผํ•˜๊ธฐ ๋•Œ๋ฌธ์— VIA ๋ณ€ํ™˜์ด ์ด๋ฃจ์–ด์ ธ๋„ ๊ฐ๊ฐ์˜ ํฌ๊ธฐ ๋ฐ ์‚ฌ์ด์ฆˆ๊ฐ€ ๋™์ผํ•จ์„ ์•Œ ์ˆ˜ ์žˆ๋‹ค.

๋‹ค์Œ์œผ๋กœ, ํ‘œ 2๋Š” EIA ์˜์ƒ์— ๋Œ€ํ•œ ์••์ถ•๋ฅ  ๋ฐ color PSNR ๊ฒฐ๊ณผ ํ‘œ์ด๋‹ค. ์••์ถ•๋ฅ ์˜ ๊ฒฝ์šฐ, high profile๋กœ ์••์ถ•ํ•œ ๊ฒฐ๊ณผ์— ๋น„ํ•ด baseline profile๋กœ ์••์ถ•ํ•œ ๊ฒฐ๊ณผ๊ฐ€ ํ›จ์”ฌ ์›”๋“ฑํ•จ์„ ์•Œ ์ˆ˜ ์žˆ๋‹ค. ๊ทธ๋Ÿฌ๋‚˜ PSNR์„ ๋น„๊ตํ–ˆ์„ ๋•Œ, baseline profile์— ๋Œ€ํ•œ PSNR์€ 30dB๋ฅผ ์ฑ„ ๋„˜์ง€ ๋ชปํ•œ๋‹ค๋Š” ์ ์—์„œ ๋ณต์› ์˜์ƒ์— ๋Œ€ํ•œ ํ™”์งˆ์ด ๋–จ์–ด์ง์„ ์•Œ ์ˆ˜ ์žˆ๋‹ค. VIA ์˜์ƒ์„ ์••์ถ•ํ•œ ๊ฒฐ๊ณผ ์—ญ์‹œ ๋น„์Šทํ•œ ๊ฒฐ๊ณผ๋ฅผ ๋ณด์—ฌ์ค€๋‹ค. ํ‘œ 4์—์„œ ๋ณด์ด๋Š” ๋ฐ”์™€ ๊ฐ™์ด, baseline profile์„ ์ ์šฉํ•˜์—ฌ ์••์ถ•ํ•œ ๊ฒฐ๊ณผ๋Š” ์••์ถ•๋ฅ ์€ ๋†’์ง€๋งŒ PSNR์ด ํ˜„์ €ํžˆ ๋–จ์–ด์ง์„ ์•Œ ์ˆ˜ ์žˆ๋‹ค.

๊ทธ๋ฆผ. 4. ์‹ค์ œ EIA ์˜์ƒ์— ๋Œ€ํ•œ VIA ๋ณ€ํ™˜ ์‹คํ—˜ ์˜์ƒ

Fig. 4. Experiment of EIA to VIA transform

../../Resources/kiee/KIEE.2020.69.1.197/fig4.png

ํ‘œ 1. ์‹คํ—˜์— ์‚ฌ์šฉ๋œ ์˜์ƒ ์ •๋ณด

Table 1. Image information for experiment

EIA

VIA

Image Size

8192 x 8192

Number of Lens

32 x 32

32 x 32

Elemental Image Size

32 x 32

32 x 32

ํ‘œ 2. EIA ์˜์ƒ์— ๋Œ€ํ•œ ์••์ถ•๋ฅ  ๋Œ€๋น„ color PSNR

Table 2. color PSNR per compression rate for EIA image

EIA Compression result

H.264/AVC high profile

H.264/AVC baseline profile

compression rate

color PSNR

compression rate

color PSNR

38.18

36.15

28.31

29.75

54.68

35.92

48.57

28.93

71.71

34.48

77.80

27.53

73.01

32.84

117.01

25.84

๋‹ค์Œ์œผ๋กœ, ๊ทธ๋ฆผ 5๋Š” ๊ฐ ์˜์ƒ์— ๋Œ€ํ•ด H.264 High profile๊ณผ Baseline profile์„ ์ ์šฉํ•œ ์••์ถ•๋ฅ  ๋Œ€๋น„ PSNR ๊ฒฐ๊ณผ์ด๋‹ค. ๋จผ์ €, ํ”„๋กœํŒŒ์ผ ๋ณ„ ์˜์ƒ์— ๋Œ€ํ•œ ๊ฒฐ๊ณผ์— ๋Œ€ํ•ด ๋ถ„์„ํ•˜๊ณ ์ž ํ•œ๋‹ค. ๊ทธ๋ฆผ 5์—์„œ ๋ณด์ด๋Š” ๋ฐ”์™€ ๊ฐ™์ด, high profile๋กœ ์••์ถ•๋œ EIA ์˜์ƒ๊ณผ VIA ์˜์ƒ์ด baseline profile์— ๋น„ํ•ด color PSNR์ด ์••์ถ•๋ฅ  70:1 ์ผ ๋•Œ EIA๋Š” ์•ฝ 7dB, VIA๋Š” ์•ฝ 8dB์ •๋„ ๋†’์Œ์„ ํ™•์ธํ•  ์ˆ˜ ์žˆ๋‹ค. ์••์ถ•๋ฅ ์€ high profile์˜ ๊ฒฝ์šฐ ์ตœ๋Œ€ ์••์ถ•๋ฅ ์ด EIA์ผ ๋•Œ ์•ฝ 75:1, VIA ์ผ ๋•Œ ์•ฝ 105:1์ž„์„ ํ™•์ธํ•  ์ˆ˜ ์žˆ๋‹ค. ๊ทธ์— ๋ฐ˜ํ•ด, baseline profile์€ high profile๋ณด๋‹ค ํ›จ์”ฌ ๋†’์€ ์•ฝ 117:1, 153:1์˜ ์••์ถ•๋ฅ ์„ ๋ณด์—ฌ์ฃผ๊ณ  ์žˆ๋‹ค. ์ด๋Š”, ๊ฐ๊ฐ์˜ ํ”„๋กœํŒŒ์ผ์ด ๊ฐ€์ง€๋Š” ์˜ค๋ฒ„ํ—ค๋“œ์˜ ํฌ๊ธฐ๋กœ ์ธํ•ด ๋†’์€ ์‚ฌ์–‘์˜ ํ”„๋กœํŒŒ์ผ์„ ์‚ฌ์šฉํ•จ์—๋„ ๋ถˆ๊ตฌํ•˜๊ณ  ์ „์ฒด ์••์ถ•๋ฅ ์ด ์ €ํ•˜๋จ์„ ๋ณด์—ฌ์ฃผ๊ณ  ์žˆ๋‹ค.

ํ‘œ 3. VIA ์˜์ƒ์— ๋Œ€ํ•œ ์••์ถ•๋ฅ  ๋Œ€๋น„ color PSNR

Table 3. color PSNR per compression rate for VIA image

VIA Compression result

H.264/AVC high profile

H.264/AVC baseline profile

compression rate

color PSNR

compression rate

color PSNR

42.01

35.45

23.86

28.05

66.56

35.17

49.66

27.41

100.96

33.22

100.41

25.99

103.26

31.85

153.41

24.37

๊ทธ๋ฆผ. 5. ํ”„๋กœํŒŒ์ผ ๋ณ„ ์••์ถ•๋ฅ  ๋Œ€๋น„ color PSNR

Fig. 5. color PSNR per compression rate corresponding to profile

../../Resources/kiee/KIEE.2020.69.1.197/fig5.png

์‹คํ—˜ ๊ฒฐ๊ณผ๋ฅผ ํ†ตํ•ด, EIA ๋ฐ์ดํ„ฐ ๋ฐ VIA ๋ฐ์ดํ„ฐ๋ฅผ ๊ณ ์••์ถ•ํ•˜๊ธฐ ์œ„ํ•ด baseline profile์„ ์‚ฌ์šฉํ•˜๊ธฐ์—๋Š” PSNR์ด ๋งค์šฐ ๋‚ฎ์•„ ํ™”์งˆ์— ๋Œ€ํ•œ ์‚ฌ์šฉ์ž์˜ ๋งŒ์กฑ๋„๋ฅผ ๋–จ์–ดํŠธ๋ฆฌ๊ณ , ํ™”์งˆ ๊ฐœ์„ ์„ ์œ„ํ•ด high profile์„ ์‚ฌ์šฉํ•˜๊ธฐ์—๋Š” ์••์ถ•๋ฅ ์ด ํ˜„์ €ํžˆ ๋–จ์–ด์ง„๋‹ค๋Š” ์ ์—์„œ ํ•œ ๋ฐฉ๋ฒ•๋งŒ์„ ๊ณ ์ˆ˜ํ•˜๊ธฐ์—๋Š” ์–ด๋ ค์›€์ด ์žˆ์—ˆ๋‹ค. ์ด๋Ÿฌํ•œ ์ด์œ ๋กœ, ์ž…๋ ฅ ์˜์ƒ์— ๋Œ€ํ•œ ํŠน์„ฑ์„ ๋ฉด๋ฐ€ํžˆ ํŒŒ์•…ํ•˜์—ฌ ์˜์ƒ์— ์ ํ•ฉํ•œ ํ”„๋กœํŒŒ์ผ ๋ฐ ์˜์ƒ ๋ณ€ํ™˜์„ ํ†ตํ•œ ์••์ถ• ๊ณผ์ •์„ ์ ์šฉํ•˜๋Š” ๊ฒƒ์ด ์ ํ•ฉํ•˜๋‹ค๋Š” ๊ฒƒ์„ ์•Œ ์ˆ˜ ์žˆ์—ˆ๋‹ค.

5. ๊ฒฐ ๋ก 

๋ณธ ๋…ผ๋ฌธ์—์„œ๋Š” ์š”์†Œ ์˜์ƒ ๋ฐฐ์—ด์„ ํšจ์œจ์ ์œผ๋กœ ์••์ถ•ํ•˜๊ธฐ ์œ„ํ•ด ์ ํ•ฉํ•œ ์˜์ƒ ๋ณ€ํ™˜ ๋ฐ ์••์ถ• ์ฝ”๋ฑ์„ ์ฐพ๊ธฐ ์œ„ํ•œ ์‹คํ—˜์„ ์ง„ํ–‰ํ•˜์˜€๋‹ค. ์ผ๋ฐ˜์ ์œผ๋กœ ๊ณ ํšจ์œจ ์••์ถ•์„ ๋ณด์—ฌ์ฃผ๋Š” H.264/AVC๋Š” ์˜์ƒ ํฌ๊ธฐ๊ฐ€ ๋งค์ฃผ ์ž‘์€ ์˜์ƒ์—๋Š” ์ ํ•ฉํ•˜์ง€ ์•Š๋‹ค๋Š” ๊ฒƒ์„ ํ™•์ธํ•  ์ˆ˜ ์žˆ์—ˆ๋‹ค. ๋˜ํ•œ, ํ”„๋กœํŒŒ์ผ ๋ณ„ ์••์ถ•๋ฅ ์— ๋”ฐ๋ผ ๊ฐ ์˜์ƒ์ด ๊ฐ€์ง€๋Š” ์˜ค๋ฒ„ํ—ค๋“œ์˜ ํฌ๊ธฐ๋กœ ์ธํ•ด ์••์ถ•๋ฅ ์— ์˜ํ–ฅ์„ ๋ผ์นœ๋‹ค๋Š” ์  ๋˜ํ•œ ํ™•์ธํ•  ์ˆ˜ ์žˆ์—ˆ๋‹ค. ์ด๋Ÿฌํ•œ ์ด์œ ๋กœ, ์ž…๋ ฅ๋˜๋Š” ์š”์†Œ ์˜์ƒ์˜ ํŠน์„ฑ์— ๋”ฐ๋ผ ์ ์‘์ ์œผ๋กœ ์š”์†Œ ์˜์ƒ ๋ฐฐ์—ด์˜ ๋ณ€ํ™˜ ๊ณผ์ •์ด ํ•„์š”ํ•˜๋ฉฐ, ์˜์ƒ ๋ฐ์ดํ„ฐ์˜ ์••์ถ•๋ฅ ์— ๋”ฐ๋ผ ์••์ถ• ์ฝ”๋ฑ์˜ ํ”„๋กœํŒŒ์ผ์„ ์ ์‘์ ์œผ๋กœ ๋ณ€๋™ํ•˜๋ฉฐ ์‚ฌ์šฉํ•ด์•ผ ํ•œ๋‹ค๋Š” ์ ์„ ์•Œ ์ˆ˜ ์žˆ์—ˆ๋‹ค.

References

1 
H. G. Kim, H. Yoo, Aug 2018, Image enhancement for computed tomography using directional interpolation for sparsely-s ampled sinogram, Elsevier Optik, Vol. 166, pp. 227-235DOI
2 
Y. Piao, M. Zhang, D. Shin, H. Yoo, 15 Aug, 2013, Three-dimensional imaging and visualization using off-axially distributed image sensing, OSA Optics Letters, Vol. 38, No. 16DOI
3 
Y. Lee, H. Yoo, Feb 2017, Low-cost 3D motion capture system using passive optical markers and monocular vision, Elsevier Optik, Vol. 130, pp. 1397-1407DOI
4 
H. Yoo, Jul. 2013(online 13 Mar. 2013), Depth extraction for 3D objects via windowing technique in computational integral imaging with a lenslet array, Elsevier Optics and Lasers in Engineering, Vol. 51, No. 7, pp. 912-915DOI
5 
J. Y. Jang, H. Yoo, Sep 2019, Computational reconstruction for three-dimensional imaging via a diffraction grating, OSA Optics Express, Vol. 27, No. 20, pp. 27820-27831DOI
6 
G. Lippmann, 1908, La photographic integrale, C.R. Acad. Sci., Vol. 146, pp. 446-451Google Search
7 
H. Yoo, J. Y. Jang, Oct 2017, Intermediate elemental image reconstruction for refocused three-dimensional images in integral imaging by convolution with ๐›ฟ-function sequences, Elsevier Optics and Lasers in Engineering, Vol. 97, pp. 93-99DOI
8 
Y. Lee, H. Yoo, Feb 2016, Three-dimensional visualization of objects in scattering medium using integral imaging and spectral analysis, Elsevier Optics and Lasers in Engineering, Vol. 77, No. 2, pp. 31-38DOI
9 
H. Yoo, D. K Shin, M. Cho, Mar. 2015 (online 19 Sep. 2014), Improved depth extraction method of 3D objects using computational integral imaging reconstruction based on multiple windowing techniques, Elsevier Optics and Lasers in Engineering, Vol. 66, No. 3, pp. 105-111DOI
10 
J. Lee, D. Shin, H. Yoo, 1 Sep. 2013, Image quality improvement in computational reconstruction of partially occluded objects using two computational integral imaging reconstruction methods, Elsevier Optics Communications, Vol. 304, pp. 96-101DOI
11 
H. Yoo, 3 Apr. 2013, Axially moving a lenslet array for high-resolution 3D images in computational integral imaging, OSA Optics Express, Vol. 21, No. 7, pp. 8876-8887DOI
12 
H. Yoo, 15 Jun. 2011, Artifact analysis and image enhancement in three dimensional computational integral imaging using smooth windowing technique, OSA Optics Letters, Vol. 36, No. 12DOI
13 
M. C. Forman, A. Aggoun, 1997, Quantisation strategies for 3D-DCT-based compression of full parallax 3D images, in Proceedings of Int. Conf. on Image Processing and Its Applications, 6th, pp. 32-35DOI
14 
R. Zaharia, A. Aggoun, M. McCormick, 2002, Adaptive 3D-DCT compression algorithm for continuous parallax 3D integral imaging, Signal Processing: Image Communication, Vol. 17, pp. 231-242DOI
15 
J. Jang, S. Yeom, B. Javidi, 2005, Compression of ray information in three-dimensional integral imaging using the Karhunen-Loeve transform, Opt. Eng., Vol. 44, pp. 127001.1-127001.10DOI
16 
H. H. Kang, D. H. Shin, E. S. Kim, July. 2008, Compression scheme of sub-images using Karhunen-Loeve transform in three-dimensional integral imaging, Optics Communications, Vol. 281, No. 14, pp. 3640-3647DOI
17 
ISO/IEC JTC1/SC29/WG1, February 2015, JPEG PLENO Abstract and Executive Summary, WG1N6922, 68th JPEG Meeting, Sydney, AustraliaGoogle Search
18 
T. Ebrahimi, S. Foessel, F. Pereira, P. Schelkens, Nov. 2016, JPEG Pleno : Toward an Efficient Representation of Visual Reality, IEEE MultiMedia, Vol. 23, No. 4, pp. 14-20DOI
19 
H. Arimoto, B. Javidi, 2001, Integral three-dimensional imaging with digital reconstruction, Optics Letters, Vol. 26, No. 3, pp. 157-159DOI
20 
T. Wiegand, G. J. Sullivan, G. Bjontegaard, A.Luthra, July. 2003, Overview of the H.264/AVC Video Coding Standard, IEEE Transactions on Circuits and Systems for Video Technology, Vol. 13, No. 7, pp. 560-576DOI
21 
A. Puri, X. Chen, A. Luthra, Oct. 2004, Video coding using the H.264/MPEG-4 AVC compression standdard, Signal Processing: Image Communication, Vol. 19, No. 9, pp. 793-849DOI
22 
G. J. Sullivan, T. Wiegand, Jan. 2005, Video Compressionโ€“From Concepts to the H.264/AVC Standard, in Proceedings of the IEEE, Vol. 93, No. 1, pp. 18-31DOI
23 
D. Marpe, H. Schwarz and T. Wiegand, July 2003, Context-based adaptive binary arithmetic coding in the H.264/AVC Video compression standard, IEEE Transactions on circuits and systems for video technology, Vol. 13, No. 7, pp. 620-363DOI
24 
S. Yeom, A. Stern, B. Javidi, 2004, Compression of 3D color integral images, Opt. Express, Vol. 12, No. 8, pp. 1632-1642DOI
25 
1994, Information Technology-Generic Coding of Moving Pictures and Associated Audio Information: Video, ISO/IEC 13818-2Google Search

์ €์ž์†Œ๊ฐœ

์˜ค์€์ฃผ (Eun-Ju Oh)
../../Resources/kiee/KIEE.2020.69.1.197/au1.png

2015~2018 : ์ƒ๋ช…๋Œ€ํ•™๊ต ๋ฏธ๋””์–ด์†Œํ”„ํŠธ์›จ์–ดํ•™๊ณผ ์ดํ•™์‚ฌ

2018~ํ˜„์žฌ : ์ƒ๋ช…๋Œ€ํ•™๊ต ๋Œ€ํ•™์› ์ปดํ“จํ„ฐ๊ณผํ•™๊ณผ ์žฌํ•™ ์ค‘

๊ด€์‹ฌ ๋ถ„์•ผ : ์˜์ƒ ์ฒ˜๋ฆฌ ๋ฐ ์••์ถ•

์œ ํ›ˆ (Hoon Yoo)
../../Resources/kiee/KIEE.2020.69.1.197/au2.png

1990~1997 : ํ•œ์–‘๋Œ€ํ•™๊ต ์ „์žํ†ต์‹ ๊ณตํ•™๊ณผ ๊ณตํ•™์‚ฌ

1997~1999 : ํ•œ์–‘๋Œ€ํ•™๊ต ๋Œ€ํ•™์› ์ „์žํ†ต์‹ ๊ณตํ•™๊ณผ ๊ณตํ•™์„์‚ฌ

1999~2003 : ํ•œ์–‘๋Œ€ํ•™๊ต ๋Œ€ํ•™์› ์ „์žํ†ต์‹ ๊ณตํ•™๊ณผ ๊ณตํ•™๋ฐ•์‚ฌ

2003~2005 : ์‚ผ์„ฑ์ „์ž ๋ฌด์„ ์‚ฌ์—…๋ถ€ ์ฑ…์ž„์—ฐ๊ตฌ์›

2005~2008 : ๋™์„œ๋Œ€ํ•™๊ต ์ปดํ“จํ„ฐ์ •๋ณด๊ณตํ•™๋ถ€ ์กฐ๊ต์ˆ˜

2008~2012 : ์ƒ๋ช…๋Œ€ํ•™๊ต ๋””์ง€ํ„ธ๋ฏธ๋””์–ดํ•™๋ถ€ ์กฐ๊ต์ˆ˜

2012~2017 : ์ƒ๋ช…๋Œ€ํ•™๊ต ๋ฏธ๋””์–ด์†Œํ”„ํŠธ์›จ์–ดํ•™๊ณผ ๋ถ€๊ต์ˆ˜

2017~ํ˜„์žฌ : ์ƒ๋ช…๋Œ€ํ•™๊ต ์œตํ•ฉ์ „์ž๊ณตํ•™๊ณผ ์ •๊ต์ˆ˜

1997.2.26. : ์ œ3ํšŒ ์‚ผ์„ฑ์ „์ž ํœด๋จผํ…Œํฌ ๋…ผ๋ฌธ์ƒ ๋Œ€ํšŒ ์ž…์ƒ

2003~2005 : ์‚ผ์„ฑ์ „์ž ์œ„์„ฑDMB ํฐ, ์ง€์ƒํŒŒDMB ํฐ ์ตœ์ดˆ ๊ฐœ๋ฐœ ๋ฐ ์–‘์‚ฐ

๊ด€์‹ฌ ๋ถ„์•ผ : ์˜์ƒ์ฒ˜๋ฆฌ, 3์ฐจ์› ์ง‘์  ์˜์ƒ ์ฒ˜๋ฆฌ