Inona no atao hoe Lossless Compression?<\/h3>\n
Mampihena ny haben’ny rakitra amin’ny alalan’ny famantarana sy ny fanafoanana ny faneriterena ara-statistika ny fanerena tsy misy fatiantoka nefa tsy manala vaovao. Rehefa decompressed, ny rakitra dia mitovy amin’ny tany am-boalohany, tsy misy very amin’ny kalitao na ny fahamarinan’ny data.<\/p>\n <\/div>\n <\/div>\n \n
Ahoana no fiasan’ny Compression Lossless<\/h3>\n
Mampiasa teknika isan-karazany hampihenana ny haben’ny rakitra ny algorithms famoretana tsy misy fatiantoka ary miantoka ny fanamboarana tonga lafatra ny angon-drakitra tany am-boalohany. Ireo fomba ireo dia mamakafaka ny lamina, ny habe ary ny rafitra ao anatin’ny angon-drakitra mba handrafetana azy io amin’ny fomba mahomby kokoa tsy ho very vaovao.<\/p>\n \n
Fanodinana lava lava (RLE)<\/h4>\n
RLE dia manolo ny filaharan’ny singa data mitovy (mihazakazaka) miaraka amin’ny sanda tokana sy isa. Ohatra, ny “AAAAAABBBCCCCC” dia lasa “6A3B5C”, izay mampihena be ny haben’ny angona misy filaharana miverimberina.<\/p>\n
Original: WWWWWWWWWWBBBWWWWWWWWWWWWBBBWWWWWWWWWW\nCompressed: 10W3B12W3B10W<\/pre>\n <\/div>\n <\/div>\n \n\n\n <\/i>\n <\/div>\nHuffman Coding<\/h4>\n
Ity teknika ity dia manome kaody miovaova-lava ho an’ny tarehin-tsoratra, miaraka amin’ny kaody fohy kokoa ho an’ny tarehintsoratra matetika kokoa. Ity fomba statistika ity dia manatsara ny kaody mifototra amin’ny fitsinjarana matetika ny toetra.<\/p>\n
\n Ohatra:<\/strong>\nFrequent character 'e': 101\nLess frequent 'z': 1010101011<\/pre>\n <\/div>\n <\/div>\n \n\n\n <\/i>\n <\/div>\nAlgorithm LZ77 & LZ78<\/h4>\n
Ireo fomba mifototra amin’ny rakibolana ireo dia manolo ny angona miverimberina miaraka amin’ny fanondroana dika iray efa misy ao amin’ny stream tsy voapotsitra. Izy ireo no fototry ny endrika malaza toa ny ZIP sy GIF.<\/p>\n
\n Ohatra:<\/strong>\nInstead of storing \"compression compression\"\nStore \"compression [pointer to earlier instance]\"<\/pre>\n <\/div>\n <\/div>\n \n\n\n <\/i>\n <\/div>\nDeflate algorithm<\/h4>\n
Ny fampifangaroana ny kaody LZ77 sy Huffman, Deflate dia manome famatrarana tsara amin’ny hafainganam-pandeha tsara. Ampiasaina amin’ny ZIP, PNG, ary HTTP compression (gzip) izy io, ka mahatonga azy io ho iray amin’ireo algorithm be mpampiasa indrindra.<\/p>\n
\n Fampiharana:<\/strong>\n\n
- ZIP archives<\/li>\n
- Sary PNG<\/li>\n
- famatrarana HTTP (gzip)<\/li>\n <\/ul>\n <\/div>\n <\/div>\n \n
\n\n <\/i>\n <\/div>\nKaody arithmetika<\/h4>\n
Ity teknika ity dia maneho hafatra ho isan’ny isa eo anelanelan’ny 0 ka hatramin’ny 1. Afaka mahatratra ny taham-pamokarana manakaiky ny fetran’ny entropy ara-teorika izy io, ka mahatonga azy io ho mahomby amin’ny karazana data sasany.<\/p>\n
\n Tombontsoa:<\/strong>\nAfaka mandika bitika fractional isaky ny marika, manome famatrarana tsara kokoa noho ny Huffman ho an’ny loharano maro.<\/p>\n <\/div>\n <\/div>\n \n
\n\n <\/i>\n <\/div>\nDelta Encoding<\/h4>\n
Raha tokony ho mitahiry sanda tanteraka, ny delta encoding dia mitahiry ny fahasamihafana eo amin’ny soatoavina mifandimby. Tena mandaitra tokoa izany ho an’ny angon-drakitra izay mitovitovy ny sanda mifanila aminy, toy ny santionany amin’ny feo na ny famakiana sensor.<\/p>\n
\n Ohatra:<\/strong>\nOriginal: 105, 107, 106, 110, 108\nDelta: 105, +2, -1, +4, -2<\/pre>\n <\/div>\n <\/div>\n <\/div>\n \nFomba fisie tsy misy fatiantoka mahazatra<\/h3>\n
\n\n\n <\/i>\nArchives<\/h4>\n <\/div>\n
\nZIP<\/div>\nRAR<\/div>\n7Z<\/div>\nGZIP<\/div>\nBZIP2<\/div>\nTAR<\/div>\n <\/div>\n <\/div>\n \n\n\n <\/i>\nSary<\/h4>\n <\/div>\n
\nPNG<\/div>\nTIFF<\/div>\nBMP<\/div>\nGIF<\/div>\nWebP (tsy very)<\/div>\n <\/div>\n <\/div>\n \n\n\n <\/i>\nAudio<\/h4>\n <\/div>\n
\nFLAC<\/div>\nALAC<\/div>\nWAV<\/div>\nAPE<\/div>\nWavPack<\/div>\n <\/div>\n <\/div>\n <\/div>\n <\/div>\n <\/section>\n \n \n\n Lossy Compression Explained<\/h2>\n \n
\n\n\n <\/i>\n <\/div>\n\nInona no atao hoe Lossy Compression?<\/h3>\n
Ny famatrarana fatiantoka dia mampihena ny haben’ny rakitra amin’ny al\u00e0lan’ny fanafoanana tanteraka ny fampahalalana sasany, indrindra fa ny angona mihoa-pampana na tsy dia manan-danja loatra. Ny rakitra decompressed dia tsy mitovy amin’ny tany am-boalohany, fa ny fahasamihafana dia natao ho sarotra na tsy ho hitan’ny olombelona amin’ny toe-javatra mahazatra.<\/p>\n <\/div>\n <\/div>\n \n
Ahoana ny fiasan’ny Lossy Compression<\/h3>\n
Ny fanerena very dia mahatratra ny tahan’ny fanerena ambony kokoa amin’ny alalan’ny fandraisana fanapahan-kevitra stratejika momba ny angona hariana. Ireo algorithm ireo dia mampiasa fahalalana momba ny fomba fijerin’ny olombelona\u2014izay azon’ny masontsika sy ny sofinantsika ary tsy hitan’ny masontsika\u2014mba hanesorana ny vaovao amin’ny fomba manamaivana ny fiantraikany miharihary eo amin’ny kalitao.<\/p>\n \n
\n\n\n <\/i>\n <\/div>\nTransform Coding<\/h4>\n
Ity teknika ity dia manova ny angona avy amin’ny sehatra iray (toy ny spatial) mankany amin’ny iray hafa (toy ny matetika) izay ahafahana mampihatra amin’ny fomba mahomby kokoa ny fanerena. Ny Discrete Cosine Transform (DCT) ampiasaina amin’ny JPEG dia ohatra tsara indrindra.<\/p>\n
\n Dingana:<\/strong>\n\n
- Ovay ho singa matetika ny sakana sary<\/li>\n
- Hamafiso amin’ny fomba mahery vaika kokoa ireo singa avo lenta<\/li>\n
- Tsy dia saro-pady loatra ny mason’ny olombelona amin’ireo matetika ireo<\/li>\n <\/ul>\n <\/div>\n <\/div>\n \n
\n\n <\/i>\n <\/div>\nQuantization<\/h4>\n
Mampihena ny fahamarinan’ny soatoavina angona ny quantization. Izy io dia mametraka sarin-tany isan-karazany amin’ny soatoavina fidirana amina andiana soatoavina kely kokoa, mampihena tsara ny isan’ny bits ilaina hanehoana ny angona.<\/p>\n
\n Ohatra:<\/strong>\nOriginal values: 4.13, 4.28, 4.97, 4.02\nQuantized to: 4, 4, 5, 4<\/pre>\n <\/div>\n <\/div>\n \n\n\n <\/i>\n <\/div>\nModeling psychoacoustic<\/h4>\n
Ampiasaina amin’ny fanerena feo, ity teknika ity dia manararaotra ny fetran’ny fandrenesan’ny olombelona. Izy io dia mamaritra ny singa audio azo esorina nefa tsy misy fiantraikany amin’ny kalitaon’ny feo.<\/p>\n
\n Hevi-dehibe:<\/strong>\n\n
- Saron-tava: Feo mafy kokoa sarona feo mangina kokoa<\/li>\n
- Fahatsapan-tena matetika: Ny fandrenesan’ny olona dia mandre matetika indrindra<\/li>\n
- Fisaronana ara-potoana: Ny feo dia afaka manarona ny hafa izay miseho fotoana fohy mialoha\/aorian’ny<\/li>\n <\/ul>\n <\/div>\n <\/div>\n \n
\n\n <\/i>\n <\/div>\nPerceptual Coding<\/h4>\n
Mitovy amin’ny modeling psychoacoustic fa ho an’ny angon-drakitra hita maso, ity fomba ity dia manaisotra ny vaovao tsy dia tsikaritry ny mason’olombelona, \u200b\u200bindrindra amin’ny antsipiriany avo lenta sy ny fiovaovan’ny loko.<\/p>\n
\n Fampiharana:<\/strong>\nAmpiasaina amin’ny JPEG, MPEG, ary ny fenitry ny famatrarana hita maso mba hanomezana laharam-pahamehana ny angona manan-danja.<\/p>\n <\/div>\n <\/div>\n \n
\n\n <\/i>\n <\/div>\nOnitra Motion<\/h4>\n
Teknika famandrihan-dahatsary izay manararaotra ny famerana ara-potoana amin’ny al\u00e0lan’ny famadihana ny fahasamihafana eo amin’ny sary fa tsy ny sary feno tsirairay. Ny fiovana avy amin’ny frame iray mankany amin’ny manaraka ihany no voakodia tanteraka.<\/p>\n
\n Dingana:<\/strong>\n\n
- Tehirizo tsindraindray ny “keyframes” (I-frames).<\/li>\n
- Ho an’ny sary hafa, tehirizo ihany ny fahasamihafana (P-frame) na ny fahasamihafan’ny bidirectional (B-frames)<\/li>\n
- Vokatry ny fampihenana ny haben’ny rakitra ho an’ny horonan-tsary<\/li>\n <\/ul>\n <\/div>\n <\/div>\n \n
\n\n <\/i>\n <\/div>\nChroma Subsampling<\/h4>\n
Ity teknika ity dia mampihena ny fampahalalam-baovao momba ny loko kokoa noho ny fampahalalana momba ny famirapiratana, manararaotra ny fahatsapan’ny mason’ny olombelona bebe kokoa amin’ny famirapiratana noho ny fahasamihafan’ny loko.<\/p>\n
\n Fomba mahazatra:<\/strong>\n\n
- 4:4:4 – Tsy misy subsampling (loko feno)<\/li>\n
- 4:2:2 – Manasasaka ny famaha ny loko marindrano<\/li>\n
- 4:2:0 – Manasasaka ny famaha ny loko mitsivalana sy mitsangana<\/li>\n <\/ul>\n <\/div>\n <\/div>\n <\/div>\n \n
Fomban-drakitra mahazatra Lossy<\/h3>\n
\n\n\n <\/i>\nSary<\/h4>\n <\/div>\n
\nJPEG<\/div>\nWebP (very)<\/div>\nJPEG 2000<\/div>\nHEIF<\/div>\nAVIF<\/div>\n <\/div>\n <\/div>\n \n\n\n <\/i>\nteny<\/h4>\n <\/div>\n
\nMP3<\/div>\nAAC<\/div>\nVorbis<\/div>\nOpus<\/div>\nWMA<\/div>\n <\/div>\n <\/div>\n \n\n\n <\/i>\nVideo<\/h4>\n <\/div>\n
\nH.264<\/div>\nH.265<\/div>\nVP9<\/div>\nAV1<\/div>\nWebM<\/div>\n <\/div>\n <\/div>\n <\/div>\n <\/div>\n <\/section>\n \n \n\n Fampiharana azo ampiharina sy tranga fampiasana<\/h2>\n \n
\n\n\n\n <\/i>\n <\/div>\nSary nomerika<\/h3>\n <\/div>\n
\nFampidirana tsy misy fatiantoka<\/h4>\n
\n
- Fitahirizana endrika RAW ho an’ny mpaka sary matihanina<\/li>\n
- Fitehirizana sary manan-danja amin’ny arisiva<\/li>\n
- Sary mila fanodinana na fanitsiana be dia be<\/li>\n
- endrika PNG ho an’ny sary misy soratra na maranitra sisiny<\/li>\n <\/ul>\n \n
Lossy Compression<\/h4>\n
\n
- JPEG ho an’ny sary isan’andro sy fizarana tranonkala<\/li>\n
- Famoronana thumbnail ho an’ny galleries sy previews<\/li>\n
- Mampakatra ny media sosialy izay mihatra ny fetran’ny habeny<\/li>\n
- Fandefasana mailaka sy fampiharana hafatra<\/li>\n <\/ul>\n \n
\n Fomba fanao tsara indrindra:<\/strong> Mak\u00e0 amin’ny endrika RAW na tsy misy fatiantoka, mamorona dikan-teny very ho an’ny fizarana, mitazona tompo tsy misy very ho an’ny tahiry.\n <\/div>\n <\/div>\n <\/div>\n \n\n\n\n <\/i>\n <\/div>\n