Nekuda kwekuwedzera kuri kuramba kuchiwedzera kwedanho rekuvandudza tekinoroji yeruzivo, gore rega rega magwaro emagetsi ari kuita nyore uye ari kudiwa mukushandiswa uye ari kutanga kutonga pamusoro pechinyakare mapepa media. Nokudaro, zvakakosha zvikuru kubhadhara panguva yakakodzera kuchengetedza zvinyorwa zvemashoko kwete chete pamapepa echinyakare, asiwo pamagwaro emagetsi. Kambani hombe yega yega ine kutengeserana, nyika uye zvimwe zvakavanzika inoda kudzivirira kuburitswa kweruzivo uye kukanganisa kweruzivo rwakasarudzika, uye kana kuvuza kwaonekwa, tora matanho ekumisa kudonha uye kuona anotyora.
Zvishoma nezvesarudzo dzekudzivirira
Kuita mabasa aya, zvimwe zvinodzivirira zvinounzwa. Zvinhu zvakadaro zvinogona kuva mabhakodhi, ma tag anooneka, ma tag emagetsi, asi anonyanya kufadza ma tag akavanzika. Mumwe wevamiriri vanonyanya kushamisa watermarks; anogona kuiswa pabepa kana kuwedzerwa asati adhindwa paprinta. Haisi chakavanzika kuti vanodhinda vanoisa yavo watermarks (yero madotsi uye mamwe mamaki) painodhinda, asi isu tichafunga nezvezvimwe zvigadzirwa zvinogona kuiswa pakombiyuta yebasa kubasa remushandi. Zvigadzirwa zvakadaro zvinogadzirwa neakakosha software package iyo inodhirowa zvigadzirwa pamusoro penzvimbo yebasa yemushandisi, ichideredza kuoneka kwezvakagadzirwa pachavo uye pasina kukanganisa basa remushandisi. Aya matekinoroji ane midzi yekare maererano nekuvandudzwa kwesainzi uye maalgorithms anoshandiswa kupa ruzivo rwakavanzika, asi haawanzo kuwanikwa munyika yanhasi. Iyi nzira inonyanya kuwanikwa munharaunda yemauto uye pamapepa, kuitira kukurumidza kuzivikanwa kwevashandi vasina kunaka. Aya matekinoroji achangotanga kuunzwa munzvimbo yekutengeserana. Anoonekwa mawatermark ave kushandiswa nesimba kuchengetedza copyright yeakasiyana midhiya mafaera, asi asingaonekwe haawanzo. Asi dzinomutsawo kufarira kukuru.
Security Artifacts
Zvisingaoneki kuvanhu Watermarks inoumba zvinhu zvakasiyana-siyana zvinogona kuva, mumutemo, zvisingaoneki kune ziso revanhu, uye zvinogona kuvharwa mumufananidzo nenzira yemaduku maduku. Tichafunga nezvezvinhu zvinooneka, sezvo izvo zvisingaoneki neziso zvingave zviri kunze kweyakajairwa nzvimbo yemavara yevazhinji monitors. Zvigadzirwa izvi zvakakosha zvakanyanya nekuda kwekukwira kwazvo kwekusaonekwa. Zvisinei, hazvibviri kuita kuti CEHs isingaoneki zvachose. Mukuita kwavo kuita, imwe mhando yekukanganiswa kwechifananidzo chemidziyo inounzwa mumufananidzo, uye imwe mhando yezvigadzirwa zvinoonekwa pairi. Ngatitarisei mhando mbiri dzezvinhu:
- Cyclical
- Chaotic (yakaunzwa nekushandurwa kwemufananidzo)
Cyclic elements inomiririra imwe magumo kutevedzana kwekudzokorora zvinhu zvinodzokororwa kanopfuura kamwe pamufananidzo wescreen (Fig. 1).
Chaotic artifacts inogona kukonzerwa nemhando dzakasiyana dzekushandurwa kwechifananidzo chakafukidzwa (Fig. 2), semuenzaniso, kuiswa kwehologram.
Mupunga. 1 Zvigadzirwa zvemabhasikoro
Mupunga. 2 Chaotic Artifacts
Kutanga, ngatitarisei sarudzo dzekuziva cyclic artifacts. Zvigadzirwa zvakadaro zvinogona kuva:
- mameseji watermark anodzokorora pachiratidziri
- binary sequences
- seti yemapoinzi echaotic mune yega grid cell
Zvese zvakanyorwa zvinyorwa zvinoiswa zvakananga pamusoro pezvakaratidzwa zvemukati; saizvozvo, ivo vanogona kucherechedzwa nekuzivisa yemuno extrema yehistogram yega yega chiteshi chemavara uye, nekudaro, kucheka mamwe ese mavara. Iyi nzira inosanganisira kushanda pamwe nekusanganiswa kwekunyanyisa kwenzvimbo kweimwe neimwe yehistogram chiteshi. Dambudziko riri pakutsvaga kwekunyanyisa kwepanzvimbo mumufananidzo wakajeka une zvakawanda zvakapinza zvekuchinja; iyo histogram inotaridzika chaizvo sawtooth, izvo zvinoita kuti nzira iyi isashande. Iwe unogona kuedza kuisa mafirita akasiyana, asi ivo vanounza yavo yekukanganisa, izvo zvinogona kupedzisira zvatungamira mukutadza kuona iyo watermark. Iko kune zvakare sarudzo yekuziva izvi zvigadzirwa uchishandisa mamwe madhigirii ekumucheto (semuenzaniso, iyo Canny edge detector). Nzira idzi dzine nzvimbo yadzo yezvakagadzirwa zvakapinza mukuchinja; madhijitari anogona kujekesa mufananidzo contours uye obva asarudza mavara emukati mukati meiyo contours kuti awedzere mufananidzo kuitira kuti uwedzere kujekesa zvigadzirwa pachazvo, asi nzira idzi dzinoda kurongeka kwakaringana kuratidza. inodiwa contours, pamwe nekuzotevera kubhinari kwechifananidzo pachacho maererano nemavara ari mumakona akasarudzwa. Aya maalgorithms anoonekwa seasina kuvimbika uye anoedza kushandisa yakagadzikana uye yakazvimirira yerudzi rwemavara ezvikamu zvemufananidzo.
Mupunga. 3 Watermark mushure mekutendeuka
Kana zviri zvemhirizhonga zvambotaurwa, ma algorithms ekuaziva achave akasiyana zvakanyanya. Sezvo kuumbwa kwechaotic artifacts kunofungidzirwa nekuisa imwe watermark pamufananidzo, iyo inoshandurwa nedzimwe shanduko (semuenzaniso, discrete Fourier shanduko). Artifacts kubva kushandurudzo dzakadaro dzakagoverwa pahwindo rese uye zvakaoma kuziva maitiro avo. Zvichienderana neizvi, iyo watermark ichave inowanikwa mukati memufananidzo muchimiro che "random" zvigadzirwa. Kuzivikanwa kwe watermark yakadaro kunouya pasi kune kutungamira shanduko yemufananidzo uchishandisa shanduko mabasa. Chigumisiro chekushandura chinoratidzwa mumufananidzo (Fig. 3).
Asi akati wandei matambudziko anomuka anodzivirira watermark kuzivikanwa mune isingasviki mamiriro akanaka. Zvichienderana nerudzi rwekutendeuka, panogona kunge paine matambudziko akasiyana, semuenzaniso, kusagoneka kwekucherechedzwa kwegwaro rinowanikwa nekutora mafoto pakona hombe zvichienderana nechidzitiro, kana kungoita foto yehurombo husina kunaka, kana skrini inochengetwa mukati. faira ine yakanyanya kurasikirwa compression. Matambudziko ese aya anotungamira mukukanganisika kwekuziva watermark; kana iri yemufananidzo wakakomberedzwa, zvinodikanwa kushandisa chero shanduko yakaoma kana kushandisa shanduko yeaffine kumufananidzo, asi haivimbisi kudzoreredzwa kwakazara kwe watermark. Kana tikafunga nezvenyaya yekubatwa kwechidzitiro, matambudziko maviri anomuka: yekutanga kukanganisa kana ichiratidza pachiratidziro pachayo, yechipiri ndeyekukanganisa kana uchichengetedza chifananidzo kubva pachiratidziro pachacho. Yekutanga yakaoma kudzora nekuda kwekuti kune matrices evatariri vemhando dzakasiyana, uye nekuda kwekushaikwa kweimwe kana imwe ruvara, ivo vanopindirana ruvara zvichienderana nekumiririra kwavo kwemavara, nekudaro vachiunza kukanganisa mu watermark pachayo. Yechipiri yakatonyanya kuoma, nekuda kwekuti iwe unogona kuchengetedza skrini mune chero fomati uye, nekudaro, kurasikirwa nechikamu chemavara emhando, saka, isu tinogona kungorasikirwa ne watermark pachayo.
Matambudziko ekuita
Munyika yanhasi, kune akawanda maalgorithms ekuunza watermark, asi hapana anovimbisa 100% mukana wekuwedzera kuzivikanwa kwewatermark mushure mekuita kwayo. Chinetso chikuru ndechekuona seti yemamiriro ekubereka anogona kubuda mune imwe neimwe nyaya. Sezvambotaurwa, zvakaoma kugadzira algorithm yekuzivikanwa iyo yaizofunga zvese zvinogoneka maficha ekukanganisa uye kuedza kukuvadza watermark. Semuyenzaniso, kana sefa yeGaussian ikaiswa pamufananidzo uripo, uye zvinhu zviri mumufananidzo wekutanga zvaive zvidiki uye zvichisiyana nekumashure kwemufananidzo, zvino zvingave zvisingaite kuzviziva, kana chikamu che watermark chicharasika. . Ngationei nyaya yemufananidzo, nehupamhi hwepamusoro huchava ne moire (Fig. 5) uye "grid" (Fig. 4). Moire inoitika nekuda kweiyo discreteness yekrini matrix uye discreteness yematrix yemidziyo yekurekodha; mune ino mamiriro, maviri mesh mifananidzo yakasimudzwa pane imwe neimwe. Iyo mesh ingango vhara zvishoma iyo watermark zvigadzirwa uye kukonzera dambudziko rekuziva; moire, zvakare, mune dzimwe watermark embedding nzira inoita kuti zvisaite kuzviziva, sezvo ichipfuura chikamu chemufananidzo ne watermark.
Mupunga. 4 Mufananidzo grid
Mupunga. 5 Moire
Kuti uwedzere chikumbaridzo chekucherekedza mawatermark, zvinodikanwa kushandisa algorithms yakavakirwa pane wega-kudzidza neural network uye mukuita basa, izvo zvinozodzidzira kuziva watermark mifananidzo. Ikozvino kune huwandu hukuru hweneural network maturusi nemasevhisi, semuenzaniso, kubva kuGoogle. Kana uchida, unogona kuwana seti yemifananidzo yereferenzi uye dzidzisa neural network kuti ione inodiwa zvigadzirwa. Iyi nzira ine mikana inovimbisa yekuzivisa kunyange yakakanganiswa zvakanyanya watermark, asi kuti uzivikanwe nekukurumidza zvinoda simba rakakura rekombuta uye nguva yakareba yekudzidziswa yekuzivikanwa chaiko.
Zvese zvinotsanangurwa zvinoita sezviri nyore, asi paukanyanya kunyura mune idzi nyaya, ndipo paunowedzera kunzwisisa kuti kuziva watermark unofanirwa kushandisa nguva yakawanda mukuita chero algorithms, uye kunyange yakawanda nguva yekuiunza kune inodiwa mukana we. tichiziva mufananidzo wega wega.
Source: www.habr.com