Algorithm na Jagorar ZTE don Matsa Matsalolin Geometry Cloud maras Rasa

Ƙayyadaddun bayanai:

• Sunan samfur: Algorithm-Masu Jagorar Mahimmanci-Spatio don Matsawar Geometry Cloud maras Asara
• Marubuta: ZHANG Huiran, DONG Zhen, WANG Mingsheng
• Buga: Disamba 2023
• DOI: 10.12142/ZTECOM.202304003

Umarnin Amfani da samfur

Gabatarwa:
An ƙera samfurin don damfara bayanan girgije mai mahimmanci, magance ƙalubalen da suka danganci ƙarfin sararin samaniya da watsa bandwidth na cibiyar sadarwa.

Babban fasali:

1. Yanayin tsinkaya wanda ya dace da intraframe da gizagizai masu tsaka-tsaki ta amfani da tsawaita matsalar mai siyar da balaguro.
2. Adaftar lissafi mai daidaitawa tare da sabuntawar mahallin cikin sauri don ingantaccen lissafin yuwuwar da sakamakon matsawa.

Matakan Amfani:

Mataki 1: Raba Point Clouds
Rarraba gizagizai na aya zuwa yadudduka naúrar tare da babban axis.

Mataki 2: Yanayin Hasashen Zane
Ƙirƙirar yanayin tsinkaya ta amfani da algorithm din mai siyar da balaguro don yin amfani da ƙarin sarari da na ɗan lokaci.

Mataki 3: Rufe Ragowa
Rubuta saura cikin rafukan rafuka ta amfani da mahallin-madaidaicin mahallin lissafi don matsawa.

FAQ:

• Tambaya: Menene mahimman fa'idodin amfani da wannan samfur?
  A: Samfurin yana ba da damar ingantacciyar matse bayanan gajimare, ta yin amfani da alaƙar sararin samaniya da na ɗan lokaci don ingantaccen sakamakon matsawa.
• Tambaya: Shin wannan samfur ɗin zai iya ɗaukar gajimare-firam guda ɗaya da ma'auni da yawa?
  A: Ee, yanayin tsinkaya ya shafi duka intraframe da gizagizai na tsaka-tsakin, yana ba da damar yanayin amfani iri-iri.

ZHANG Huiran, DONG Zhen, WANG Mingsheng

1. Cibiyar Nazarin Tsare-tsare da Zane-zane ta Guangzhou, Guangzhou 510060, Sin;
2. Babban dakin gwaje-gwaje na Kasuwancin Guangdong don fahimtar birane, sa ido da faɗakarwa na farko, Guangzhou 510060, China;
3. Mabudin dakin gwaje-gwaje na Injiniyan Watsa Labarai na Jiha a Taswirar Taswira⁃ ping da Sensing na nesa, Jami'ar Wuhan, Wuhan 430079, China)

Takaitawa: Matsawar gajimare mai nuni yana da mahimmanci don ƙaddamar da wakilcin 3D na duniyar zahiri kamar su 3D immersive telepresence, tuƙi mai cin gashin kansa, da al'adun gargajiya.tage kiyayewa. Koyaya, ana rarraba bayanan gajimare ba bisa ka'ida ba kuma ba tare da tsayawa ba a cikin sarari da yanki na wucin gadi, inda ƙarancin voxels marasa ƙarfi da ƙarancin alaƙa a cikin sararin 3D yana sa samun ingantaccen matsawa matsala ce mai wahala. A cikin wannan takarda, muna ba da shawarar algorithm mai shiryarwa na lokaci-lokaci-lokaci don matsewar geometry na girgije mara asara. Tsarin da aka tsara yana farawa tare da rarraba gajimare mai ma'ana zuwa yankakken yadudduka na kauri mai tsayi tare da mafi tsayi. Sa'an nan kuma, yana gabatar da hanyar tsinkaya inda duka intraframe da gizagizai masu tsaka-tsakin suna samuwa, ta hanyar ƙayyade ma'auni tsakanin yadudduka da ke kusa da kuma kimanta hanya mafi guntu ta amfani da algorithm mai sayarwa mai tafiya. A ƙarshe, ƴan ragowar tsinkayar an matse su da inganci tare da ingantacciyar hanyar mahallin jagora da dabarun ƙididdiga masu saurin daidaitawa. Gwaje-gwaje sun tabbatar da cewa hanyar da aka tsara za ta iya samun nasarar cimma ƙarancin ƙarancin ƙarancin rashi na bayanan geometric na ma'ana girgije, kuma ya dace da matsawar girgije mai ma'ana ta 3D wanda ya dace da nau'ikan al'amuran daban-daban.
Mahimman kalmomi: ma'anar matsi na lissafi na girgije; gizagizai masu ma'ana guda ɗaya; gajimare mai ma'ana da yawa; coding na tsinkaya; coding lissafi.

Bayanin (Tsarin 1): ZHANG HR, DONG Z, WANG M S. Spatio-lokaci-lokaci mahallin algorithm shiryar don rashin hasarar ma'ana gajimare geometry matsawa [J]. Sadarwar ZTE, 2023, 21 (4): 17-28. DOI: 10.12142/ZTECOM.202304003
Bayanin (Tsarin 2): HR. 21, ba. 4, shafi 17-28, Dec. 2023. doi: 10.12142/ZTECOM.202304003.

 

Gabatarwa

Tare da haɓakar dandamali da yawa da aikin sayan kayan aiki da yawa, fasahar gano haske da kewayon (LiDAR) na iya daidaita abubuwa na 3D da kyau tare da manyan saiti. Idan aka kwatanta da bayanan multimedia na gargajiya, bayanan girgije na nuni sun ƙunshi ƙarin bayanan ma'aunin jiki wanda ke wakiltar abubuwa daga kyauta viewmaki, har ma da al'amuran da ke da hadadden tsarin topological. Wannan yana haifar da tasirin ma'amala mai ƙarfi da haɓakawa waɗanda ke ba masu amfani da ƙwarewar gani da gaske. Bugu da ƙari, bayanan girgije na nuni suna da ƙarfin hana surutu da ƙarfin aiki iri ɗaya, wanda da alama ya sami jan hankali daga masana'antu da ilimi, musamman ga wuraren aikace-aikace kamar al'adun gargajiya.tage kiyayewa, 3D immersive telepresence da tuki ta atomatik[1-2].
Koyaya, bayanan girgije na nuni yawanci suna ƙunshe da miliyoyi zuwa biliyoyin maki a cikin wuraren sararin samaniya, suna kawo nauyi da ƙalubale ga iyawar sararin ajiya da bandwidth watsa hanyar sadarwa. Misali, girgije mai ƙarfi gama gari da ake amfani da shi don nishaɗi yakan ƙunshi kusan maki miliyan ɗaya a kowane firam, wanda, a firam 30 a sakan daya, ya kai jimlar bandwidth na 3.6 Gbit/s idan an bar shi ba a matsawa[3]. Sabili da haka, bincike akan ingantaccen matsi na juzu'i na juzu'i don gizagizai na ma'ana yana da mahimmancin ƙima da ƙima.
Aikin farko ya magance wannan matsala ta hanyar gina grid kai tsaye ko kan buƙatun ƙasaampling, saboda iyakancewa a cikin ikon sarrafa kwamfuta da ƙimar tattarawar gajimare, wanda ya haifar da ƙarancin aikin matsawa na lokaci-lokaci da asarar bayanan fasalin sifa na geometric. Nazarin kwanan nan ya dogara ne akan zane-zanen kwamfuta da dabarun sarrafa siginar dijital don aiwatar da toshe ayyukan akan bayanan girgije [4 5] ko haɗin fasahar coding bidiyo [6 7] don ingantawa. A cikin 2017, Ƙungiyar Ƙwararrun Hotunan Motsi (MPEG) ta nemi shawarwari don matsawar girgije da kuma gudanar da tattaunawa ta gaba kan yadda za a damfara irin wannan bayanan. Tare da haɓaka hanyoyin da za a nuna ma'aunin girgije da aka samo da kuma gabatar da su, an ba da tsarin ma'auni na bayanai na girgije guda biyu-TMC13 da TMC2 a cikin 2018. Binciken da ke sama ya nuna ci gaba mai ban mamaki a cikin fasahar matsawa na ma'ana girgije. Koyaya, aikin da ya gabata galibi yana hulɗa da alaƙar sararin samaniya da na ɗan lokaci na gajimare daban amma har yanzu ba a yi amfani da su ga cikakkiyar damarsu ta matsewar girgije ba.
Don magance ƙalubalen da aka ambata a baya, mun gabatar da hanyar da za ta jagoranci mahallin lokaci-lokaci don matsewar lissafi na girgije mara asara. Da farko za mu raba gizagizai masu ma'ana zuwa jeri na raka'a tare da babban axis. Sannan muna ƙirƙira yanayin tsinkaya ta hanyar algorithm mai siyarwa mai balaguro, ta hanyar ɗaukar alaƙar lokaci. A ƙarshe, ragowar abubuwan an rubuta su cikin rafukan rafukan da aka yi amfani da su na mahallin-madaidaicin ƙididdiga. Babban gudunmawar mu shine kamar haka.
1) Mun ƙirƙira- yanayin tsinkaya wanda ya dace da duka intra-frame da girgije mai tsaka-tsakin tsaka-tsakin, ta hanyar tsawaita matsalar mai siyar da balaguro (TSP). Ta hanyar yin amfani da abubuwan da suka shafi sararin samaniya da na ɗan lokaci na gizagizai, hasashen lissafi na iya yin amfani da alaƙar sararin samaniya don haka yana ba da damar nau'ikan yanayi iri-iri.
2) Muna gabatar da mai rikodin lissafin daidaitacce tare da sabuntawar rubutu cikin sauri, wanda ke zaɓar mahallin 3D mafi kyau daga ƙamus na mahallin, kuma yana hana haɓakar ƙimar entropy. Sakamakon haka, yana haɓaka ƙimar yuwuwar lissafin ƙididdiga na entropy kuma yana haifar da sakamako mai mahimmanci.
Sauran wannan takarda an tsara su kamar haka. Sashi na 2 yana ba da fayyace aikin da ke da alaƙa akan matsewar geometry na aya. Sashi na 3 da farko yana gabatar da ƙarewaview na tsarin da aka tsara. Sa'an nan kuma, hanyar da aka tsara za a bayyana dalla-dalla. An gabatar da sakamakon gwaji da ƙarshe a Sashe na 4 da 5, bi da bi.

Aiki masu alaƙa

An sami ma'ana da yawa na juzu'i na matsawa gajimare a cikin wallafe-wallafen. CAO et al. [8] da GRAZIOSI et al. [9] gudanar da bincike da taƙaita hanyoyin damtse gajimare na halin yanzu, mai da hankali kan fasahar matsawa sararin samaniya da tsarin daidaita daidaitattun MPEG bi da bi. Mun bayar da taƙaitaccen review na abubuwan da suka faru na kwanan nan a cikin nau'i biyu: matsawar girgije mai ma'ana guda-frame da matsa lamba mai yawa.

1. Matsa Matsakaicin-Frame Point Cloud
   Ana amfani da gajimare mai-firam guda ɗaya a cikin binciken injiniya, al'adun gargajiyatage adanawa, tsarin bayanan yanki, da sauran al'amura. Octree tsarin bayanai ne da ake amfani da shi sosai don wakiltar gizagizai masu inganci, waɗanda za a iya matsawa ta hanyar yin rikodin bayanai ta cikin nodes ɗin da aka mamaye. HUANG et al.[10] ba da shawarar hanyar tushen octree wanda ke rarraba ma'anar gajimare akai-akai zuwa nodes tare da matsayinsu da ke wakilta ta tsakiyar jumhuriyar kowace raka'a. FAN et al.[11] kara inganta wannan hanyar ta hanyar gabatar da bincike na cluster don samar da matakin daki-daki (LOD) matsayi da kuma sanya shi cikin tsari na farko-fadi. Duk da haka, waɗannan hanyoyin na iya haifar da ɓarna saboda kusantar ƙirar asali yayin aikin maimaitawa.
   Don magance waɗannan iyakoki, malamai sun gabatar da fasalulluka na tsarin geometric, kamar - samfurin saman triangular[12], ƙirar shimfidar wuri [13 14], da clustering al-gorithm[15], don tsinkayar tsaka-tsaki da ragowar lissafi. . RENTE et al.[16] ba da shawarar ra'ayi na matsawa mai layi na ci gaba wanda ya fara amfani da tsarin octree don ɓoye-ƙarfe-ƙarfe sannan yana amfani da jadawali Fourier don matsawa da sake gina bayanan girgije. A cikin 2019, MPEG ya fitar da fasaha na tushen juzu'i na matsawa gajimare (G-PCC) don duka a tsaye da gajimare mai tsauri, wanda ake aiwatar da shi ta hanyar daidaita sauyi, voxelization, nazarin tsarin geometric, da ƙididdigar lissafi mataki-mataki[17].
   Tun da wasu octants a cikin octree na iya zama marasa yawan jama'a ko ma fanko, an ba da shawarar wasu hanyoyin don inganta tsarin bishiyar ta hanyar yankan ƙananan nodes don haka kiyaye rabon ƙwaƙwalwar ajiya. Domin misaliampda, DRICOT et al. [18] ya ba da shawarar yanayin ƙididdigewa kai tsaye (IDCM) don ƙare ɓangaren octree bisa ƙayyadaddun sharuddan bincike na ɓarna, wanda ya haɗa da datsa tsarin octree don adana ragowa da aka ware wa nodes na yara. ZHANG et al. [19] yana ba da shawarar rarraba sararin sararin samaniya tare da manyan abubuwan haɗin gwiwa da daidaita hanyar rarraba daga bishiyar binaryar, quadtree da octree. Idan aka kwatanta da na gargajiya octree partitioning, da matasan model da aka ambata a sama yadda ya kamata rage adadin ragowa amfani da su wakilci sparn maki, saboda haka ajiye nodes da bukatar a encoded. Koyaya, ana buƙatar ƙayyadaddun yanayin hyperparameter da ƙayyadaddun yanayin a cikin tsari, yana sa ya zama da wahala a cika buƙatun daidaitawar kai da ƙarancin rikitarwa.
   Tare da zurfin hanyoyin sadarwa na jijiyoyi suna samun ci gaba mai mahimmanci a cikin hoto da matsawar bidiyo, masu bincike sun binciko hanyoyin da za a ƙara rage ƙimar kuɗi ta hanyar yin amfani da babban jagorar da ta gabata da kuma sake bayyana bayanan sararin samaniya a yayin aiwatar da matsawa. QUACH et al.[20] da HUANG et al.[21] ba da shawarar hanyoyin da suka haɗa waɗannan ra'ayoyin. GUARDA et al. Haɗa hanyoyin sadarwa na juzu'i na jijiyoyi da autoencoders don yin amfani da redundancy tsakanin maki kusa da haɓaka daidaitawar coding a cikin Ref. [22]. Kwanan nan, WANG et al. [23] ya ba da shawarar hanyar matsawa gajimare dangane da bambance-bambancen mai rikodin auto, wanda ke inganta ƙimar matsawa ta hanyar koyon hyperprior da rage yawan ƙwaƙwalwar ajiyar ƙididdiga. Hanyoyin da aka ambata a baya suna amfani da maƙallan cibiyar sadarwar jijiyoyi don ɗaukar babban tsari na ɓoye vector na gajimaren batu, yuwuwar ƙirar ƙirar entropy, da yuwuwar gefen wanda ya fi dacewa, don haka rage yawan ƙwaƙwalwar ajiya na coding lissafi. Gabaɗaya magana, bincike kan matsewar giza-gizan gizagizai mai ma'ana guda ɗaya ya balaga, amma akwai ƙalubale guda biyu da suka rage tukuna. Ba a yi amfani da alaƙar sararin samaniya yadda ya kamata ba, kuma yawancin hanyoyin ba su ƙididdige alaƙar bayanan girgije da kyau sosai da inganci. Bayan haka, lissafin ƙirar yuwuwar don coding entropy yana bayyana tsayi kuma mai wahala saboda ɗimbin adadin mahallin.
2. Multi-Frame Point Cloud Compression
   Ana amfani da gizagizai da yawa a cikin al'amuran kamar su telepresence 3D immersive telepresence, VR m, 3D kyauta. viewbatu watsa shirye-shirye da kuma atomatik tuki. Ba kamar matsawar girgije mai ma'ana guda ɗaya ba, matsa lamba mai ma'ana da yawa yana ba da fifikon amfani da daidaitawar lokaci, kazalika da kimanta motsi da ramuwa. Hanyoyin da ake da su don matsawa gajimare mai ma'ana da yawa za a iya raba su zuwa rukuni biyu: tsinkayar 2D da kayan ado na 3D.
   Fannin hoto da damfara bidiyo yana da yawa kuma an bincika sosai cikin ƴan shekarun da suka gabata. Algorithms daban-daban suna canza gajimare mai nuni zuwa hotuna sannan kuma damfara su kai tsaye ta hanyar FFmpeg da H. 265 encoders, da sauransu. AINALA et al[24] suna gabatar da tsarin tsinkayar tsari mai kama da yanayin en-coding wanda ke ɓoye duka geometry da halayen launi ta hanyar binciken raster akan jirgin sama. . Koyaya, wannan hanyar tana haifar da canje-canje a cikin sifar da aka yi niyya yayin aiwatar da taswira, yana sa ingantacciyar tsinkaya ta wahala. Saboda haka, SCHWARZ et al.[25] da SEVOM et al.[26] ba da shawarar jujjuya tsinkayar tsari, tsinkayar cube, da hanyoyin tsinkayar tushen faci don juyar da gizagizai zuwa bidiyoyin 2D, bi da bi. Ta hanyar sanya irin wannan tsinkaya a cikin firam ɗin da ke kusa a wuri ɗaya a cikin hotuna masu kusa, damfarar bidiyo na iya cire haɗin ɗan lokaci gabaɗaya. A cikin Ref. [27]. MPEG ya fito da fasaha na tushen bidiyo na matsawar girgije (V-PCC) don gajimare mai ƙarfi a cikin 2019[28]. Wannan tsarin yana raba gajimare ma'aunin shigarwa zuwa ƙananan tubalan masu kamanceceniya na al'ada da kuma ci gaba da sarari, sa'an nan kuma ya juyar da su zuwa saman shimfidar wuri ta hanyar cubes don yin rikodin hoton zama da bayanan taimako. Duk hotunan da aka samu an matsa su ta manyan codecs na bidiyo, kuma duk raƙuman raƙuman ruwa an haɗa su cikin fitarwa ɗaya. file. An yi wasu yunƙuri don inganta tasirin waɗannan hanyoyin. COSTA da al.[29] yi amfani da sabbin dabarun faci da yawa daga fuskar ingantawa don marufi algorithm, hanyoyin tattara bayanai, rarrabuwa masu alaƙa, da masu nuna matsayi. Bugu da ƙari, PARK et al. [30] ƙirƙira hanyar tattara bayanai-daidaitacce wanda zai daidaita ƙungiyoyin firam ɗin da ke kusa da su cikin rukuni ɗaya bisa ga kamannin tsarin ba tare da shafar aikin rafin V-PCC ba. Sakamakon asarar bayanai da babu makawa sakamakon tsinkayar gajimare na aya, malamai sun ɓullo da ingantattun dabaru don damfara jerin ginshiƙai na firam ɗin jere ta amfani da fasahar ramuwa ta motsi dangane da sararin samaniya na 3D. KAMMERL et al.[31] ba da shawarar hanyar rufaffiyar geometric na tushen octree, wanda ke samun ingantaccen matsi ta hanyar aiwatar da keɓancewar OR (XOR) tsakanin firam ɗin da ke kusa. Ba wai kawai an karɓi wannan hanyar a cikin sanannen ɗakin karatu na Point Cloud (PCL) [32] ba amma kuma ana amfani da shi sosai don ƙarin bincike na algorithm. Sauran hanyoyin shiga tsakani suna canza matsalar kimanta motsin motsi na 3D zuwa matsalar da ta dace da fasalin[33] ko amfani da bayanan geometric da aka sake ginawa[34] don tsinkayar abubuwan motsi da gano alaƙar da ke tsakanin firam ɗin kusa daidai. Binciken fashewa na baya-bayan nan [35 36] ya nuna cewa damfara bidiyo da aka koyo yana ba da mafi kyawun aiki-hargitsi a kan na gargajiya, yana kawo mahimmancin tunani don nuna matsawar girgije. ZHAO et al.[37] gabatar da hanyar sadarwar tsinkayar tsaka-tsakin madaukai guda biyu don aiwatar da tsinkayar tsaka-tsaki da kawo ingantaccen amfani da bayanan da suka dace a cikin sararin sarari da na ɗan lokaci. KAYA et al. [38] ƙirƙira wani sabon tsari don ɓoye fasalulluka na juzu'i na jeri-nauyen gajimare masu yawa, inganta CNN don ƙididdige rarraba rikodin don gane matsewar giza-gizai mara nauyi.
   Duk da ci gaba a cikin fasahar coding na matsawa na nau'ikan girgije mai ma'ana da yawa, matsaloli biyu sun ci gaba. Hannun hanyoyin damfarar gajimare da ke akwai da yawa sun dogara ne akan rikodin bidiyo da ramuwar motsi, wanda babu makawa ya haɗa da asarar bayanai ko murdiya sakamakon taswira da toshe ƙarshen. Bugu da ƙari, ƙididdige ƙididdiga yana nuna ƙarancin aiki saboda rashin daidaituwar jumlolin gizagizai na tsaka-tsaki. Bayyanar ɓangarorin maki tsakanin firam ɗin da hayaniyar da ba za a iya gujewa ba yana ƙara wahalar yin amfani da ƙididdiga na ƙididdiga yadda ya kamata a cikin matsi tsakanin firam ɗin.

Ƙaddamar da Ƙirar Tsari-Tsarin Mahimmanci-Jagorar Rashin Geometry Marasa Hanyar Matsawa Gajimare

Ƙarsheview
An nuna gabaɗayan bututun mu na sararin samaniya-yanayin mahallin mahallin jagora a cikin siffa. 1. Na farko, muna aiwatar da girgije mai shigar da bayanai ta hanyar amfani da voxelization da canjin sikelin. Sa'an nan, batu gajimare ya kasu kashi kashi-kauri yankakken yadudduka tare da babban axis. Bayan haka, muna tsara yanayin tsinkaya wanda ke yin cikakken amfani da bayanan daidaitawa na ɗan lokaci da na sararin samaniya a cikin firam ɗin intra-frame da tsaka-tsaki. Muna ƙididdige mafi guntu hanya na maki na nunin yadudduka (R-Layer) ta hanyar algorithms masu siyarwar tafiya, sannan ana amfani da sakamakon R-Layer don tsinkayar sararin samaniya da ɓoye sauran gajimare ma'ana, wato yadudduka annabta (P-Layer). ). A ƙarshe, ana ɗaukar ingantattun entropy codeing algorithms don samun matsi na binary file.

Hotunan Yanke-Tsakanin Matsayin Matsayi

1. Pre-processing
   Kayan aikin riga-kafi ya ƙunshi voxelization da canjin sikelin, don ingantacciyar ƙididdiga ta kowane takamaiman batu. A cikin voxelization, muna raba sararin samaniya zuwa cubes na girman N, wanda ya dace da ainihin ƙuduri na girgije mai ma'ana. Kowane batu an ba shi nau'in voxel na musamman dangane da matsayinsa. Ana yin rikodin voxel azaman 1; idan an shagaltar da shi, yana da 0 in ba haka ba. Canjin sikelin na iya rage ɓacin rai don ingantacciyar matsawa ta hanyar zuƙowa gajimare, inda nisa tsakanin maki ke ƙarami. Muna tara ma'aunin daidaitawar girgije (x, y, z) ta amfani da ma'aunin sikeli s, watau,
   Don tabbatar da matsi mara asara, muna buƙatar tabbatar da cewa ma'aunin sikelin s ba zai iya haifar da asarar lissafi ba kuma yana buƙatar yin rikodi a cikin taken. file.
2. Yankakken-Layer rabo
   Wannan tsarin yana aiki ta hanyar rarraba gajimare mai nuni na 3D tare da ɗayan gaturansa, ƙirƙirar yadudduka da aka yanka da yawa tare da bayanan da ba a shagaltar da su ba kawai waɗanda za a iya ƙara matsawa ta amfani da maƙallan tsinkaya da coder na lissafi. An bayyana aikin a matsayin:
   inda G ke nufin matrix mai daidaita ma'aunin bayanai, axis yana nufin girman da aka zaɓa, kuma S (a, b) shine yanki na 2D da kowane Layer ya fitar. Gabaɗaya, muna gudanar da gwaje-gwaje akan jerin gwaje-gwaje masu yawa, kuma sakamakon yana nuna cewa rarrabuwa tare da mafi tsayi axis na bambancin sararin samaniya yana haifar da mafi ƙarancin bitrate, watau.
3. Mafi qarancin hakar akwatin
   A mafi yawan lokuta, voxels da aka shagaltar da su yawanci ba za a iya kaucewa ba kuma sun fi yawan abin da aka shagaltar da su. Sakamakon haka, sarrafawa da ɓoye nau'ikan nau'ikan voxels guda biyu a lokaci guda suna ɗaukar nauyin haɗaɗɗen ƙididdiga da saurin ɓoyewar algorithm matsawa. Don haka, mun ɗauki akwatin da aka ɗaure kai tsaye (OBB) [39] don ƙididdige ƙaramar akwatin ɗaure ga kowane yanki mai yankakken, tabbatar da cewa kwatance kwalayen da aka ɗaure sun yi daidai da yadudduka. A cikin aiki na gaba, voxels kawai waɗanda ke cikin ƙayyadaddun ƙayyadaddun murabba'i ne aka matsa.

Rufaffen Hasashen Yanar Gizo Mai Jagora

Manufar rufaffen ɓoyayyiyar tsinkayar mahallin sararin samaniya shine a rufaffen duk maki Layer Layer. Ƙaddamar da TSP, mun ƙirƙira yanayin tsinkaya don bincika yuwuwar umarni da alaƙa a cikin kowane yanki mai yanki. Wannan tsarin ya ƙunshi bangare da lissafin hanya mafi guntu.
Da farko, muna raba sassan da aka yanka da kuma ƙayyade R-Layer da R-Layer ga kowane rukuni. Muna ratsa ma'anar ma'aunin girgije ta Layer tare da zaɓaɓɓen axis. Lokacin da tsayin babban shugabanci na ƙaramar akwatin daɗaɗɗa tsakanin yadudduka masu kusa ya bambanta ta ƙayyadadden tsayin raka'a, ana rubuta shi azaman rukuni ɗaya. In ba haka ba, ana amfani da shi azaman ma'aunin tunani na rukuni na gaba, kuma kowane batu girgije a cikin rukuni mai zuwa yana amfani da hanya mafi guntu. A cikin wannan takarda, mun saita Layer na farko na kowane rukuni a matsayin R-Layer, da sauran a matsayin P-Layer. Har ila yau, muna gudanar da gwaje-gwaje a kan adadi mai yawa na jerin gwaje-gwaje kuma muna ba da shawarar saita wannan ƙayyadaddun siga a matsayin raka'a 3 don samun mafi kyawun matsawa.
Bayan haka, muna gudanar da lissafin hanya mafi guntu akan R-layers kuma muna yin rikodin ragowar 'yan wasa. Dangane da ka'idar rarraba ma'aunin girgije na kowane yanki yanki, da kyau muna shirya gajimare marasa tsari don kowane yanki yanki bisa ga TSP algorithm. Wannan yana ba mu damar ƙididdige hanya mafi guntu da kyau zuwa ga ma'aunin girgije na R-layers, sa'an nan kuma rikodin ragowar matakan tsinkaya masu dacewa. Algorithm 1 yana nuna lambar ƙididdiga ta hanyar tsinkaya.

Da fari dai, muna ayyana ƙa'idar lissafin nisa tsakanin maki a cikin yanki na gida kuma mu fara yanayin hanyar tare da zaɓaɓɓen ma'ana pc1. A cikin kowane juzu'i, duk lokacin da aka ƙara sabon ma'ana pci, ana sabunta ruɓi ta hanyar hanyar daidaitawa ta jiha (P - i, i) har sai an rubuta duk ƙarin maki a cikin P cikin tsari mafi guntuwar hanya. Ana gyara wannan tsari a hankali bisa mafi ƙarancin ma'aunin nesa. Bayan an kammala duk maimaitawa a cikin jimillar hanya mafi guntu, muna lissafin min dist(pci, pcj) a cikin kowane R-Layer, da kuma mayar da mafi guntu hanya rikodin tebur na batu girgije a cikin kowane R-yadudduka. Don ƙarin matsawa, muna ƙididdige ɓarna na P-Layer daga mafi guntun hanyar R-Layer a cikin rukuni ɗaya kuma muna rikodin su azaman ragowar tsinkaya. A ƙarshe, mafi guntuwar hanyar Rlayer da ragowar kowane rukuni ana fitarwa kuma an wuce su zuwa entropy encoder don damfara ragowar hasashen gaba.

Hasashen Hasashen-Tsarin Rubutun Hasashen-Jagora Mai Kyau
Yanayin tsinkaya mai jagorar mahallin sararin samaniya yana ɓoye
Gizagizai masu fa'ida guda ɗaya daban-daban. Koyaya, yin amfani da bayanan sararin samaniya zuwa kowane gajimare-firam guda daban na iya rasa damar da aka fallasa ta hanyar alaƙar ɗan lokaci a cikin gajimare mai fa'ida da yawa. Yin la'akari da cewa gajimare mai fa'ida da yawa yana raba ɗimbin gizagizai, muna mai da hankali kan yin amfani da jan ƙarfe na ɗan lokaci don ƙara haɓaka ingancin matsi. Don haka, dangane da yanayin hasashen yanayi mai jagorar sararin samaniya, za mu iya damfara ma'aunin gizagizai ta hanyar gano ma'amala tsakanin yadudduka kusa da firam.

1. Inter-frame bangare
   Don haɓaka tasirin yanayin tsinkayar tsaka-tsaki, yana da mahimmanci don tabbatar da isasshiyar kamanceceniya tsakanin firam ɗin kusa. Sakamakon haka, muna buƙatar raba ƙungiyoyi tsakanin firam ɗin da ke kusa kuma mu ƙayyade R-Layer da P-Layer a cikin firam ɗin. Ta hanyar ƙididdige hanya mafi guntu na P-Layer bisa ga mafi guntun hanyar R-Layer, muna yin rikodin ragowar tsinkaya kuma muna kara matsa su ta hanyar entropy encoder. Algorithm 2 yana nuna lambar ɓoyayyiyar ɓangarori na interframe.
   Dangane da jeri-jeri-yadudduka daidaitacce, mun gane m bangare da lafiya bangare jere. Don ɓangarorin ƙaƙƙarfan, muna rarrabuwa yankan yadudduka na kowane firam dangane da daidaitawar da suka dace da gatura mai rarraba, daga ƙarami zuwa babba. Sakamakon haka, kowane yanki na kowane firam yana da lamba ta musamman, yana ba mu damar raba sassan yanki da lamba ɗaya tsakanin firam ɗin da ke kusa. Bayan haka, muna ƙididdige bambanci tsakanin babban tsayin axis na ƙananan akwatunan ɗaure na yadudduka masu kusa da lamba ɗaya. Idan wannan ƙimar ta yi ƙasa da ko daidai da ƙayyadadden naúrar tsayi, za a raba sassan zuwa rukuni ɗaya. In ba haka ba, muna kwatanta bambanci a cikin tsayin babban axis na ƙaramin akwatin ɗaure a cikin madaidaicin madaidaicin firam ɗin da ke kusa da ƙayyadadden Layer kafin da bayan lamba a cikin firam ɗin da ke kusa. Sa'an nan kuma an raba Layer tare da ƙaramin bambanci zuwa rukuni ɗaya. Wannan yana tabbatar da kyakkyawan rarrabuwa tsakanin yadudduka da ke kusa, da kuma yadda za a gane kyakkyawan ɓangaren dangantakar da ke kusa.
2. Yanayin tsinkaya mai jagorar mahallin lokaci-lokaci
   Dangane da bangare, muna amfani da fadada yanayin hasashen da aka ambata a cikin Sashe na 3.3. Muna haɗa mahallin tsaka-tsaki a cikin tsarin, ma'ana cewa farkon Layer na kowace ƙungiya, wanda ke aiki azaman R-Layer, maiyuwa ba lallai ba ne ya samar da mafi kyawun sakamakon tsinkaya. Don cikakken bincika yuwuwar alaƙa tsakanin yadudduka da ke kusa, muna buƙatar fallasa mafi kyawun yanayin tsinkaya.
   Da farko, muna ƙididdige ragowar tsinkaya don kowane yanki mai yankakken a cikin rukuni na yanzu lokacin amfani da shi azaman R-Layer. Ta hanyar kwatanta ragowar tsinkaya a kowane yanayi, muna zaɓar R-Layer tare da mafi ƙarancin ƙimar saura a matsayin mafi kyawun yanayin tsinkaya. Don lissafin mafi ƙarancin hanya na R-Layer, muna amfani da algorithm mai siyar da balaguro don ƙididdige hanya mafi guntu na R-Layer ƙarƙashin mafi kyawun yanayin tsinkaya. Bugu da ƙari, muna ƙididdige ragowar tsinkaya ga kowane rukuni a ƙarƙashin mafi kyawun yanayin hasashen su. Muna kuma yin rikodin tsawon zama da bayanan R-Layer na kowane rukuni don ƙarin matsawa a aiki na gaba. A cikin aikin na gaba, muna amfani da ƙididdiga na ƙididdiga bisa mafi kyawun zaɓin mahallin don bayanin da ke sama don kammala dukkan aiwatar da ma'aunin ma'aunin gizagizai na gizagizai masu yawa.

Rubutun Ƙirar lissafi Bisa ga ƙamus na Ma'ana
Babban adadin mahallin mahallin cikin ma'ana gajimare yana da matukar nauyi ga tsarin matsawa gabaɗaya dangane da ƙaƙƙarfan ƙididdige ƙididdigan lissafi. Muna inganta ƙididdigan ƙididdiga daga abubuwa biyu masu zuwa. 1) Mun kafa ƙamus na mahallin, kuma zaɓi da sabunta ƙimar mafi kyau ta duniya bisa ga kimanta entropy, sannan 2) muna ɗaukar maɓalli masu daidaitawa don ƙididdige manyan iyakoki na sama da ƙasa da kyau.

1. Ginin ƙamus na yanayi
2. Muna gina ƙamus na mahallin da ke wakiltar jerin gwano sau uku, wanda ya ƙunshi daidaitawar gajimaren batu a kowane yanki da aka yanka da maƙasudin maƙasudin mahallinsa mara amfani. Don haka, muna danganta voxels ɗin da ke ƙunshe a cikin gajimare mai ma'ana tare da ƙaramin akwati na ɗaure kowane Layer tare da mahallinsa mara wofi. Don kwatanta ginin ƙamus ɗin mahallin a sarari, mun ba da bayani mai zurfi a cikin siffa 2. Don murabba'i biyu masu inuwa a cikin siffa 2, wuraren taswirar mahallin pc1 da pc2 kawai ana la'akari da su. Gudunmawar mahallin tare da axis x da y-axis an rubuta su zuwa jerin layi biyu QX - da QY - bi da bi. Don haka ƙamus ɗin mahallin ya ƙunshi QX - da QY -. Abubuwan jerin gwano tare da daidaitawa iri ɗaya an haɗa su cikin nau'i-nau'i uku, wanda aka ƙididdige wakilcin mahallin mahallin a matsayin jimillar gudunmawar mahallin mahallin uku-uku.
   Don haka, ana iya ƙididdige mahallin kowane voxel azaman jimillar gudummawar masu zaman kansu na voxels da aka mamaye a cikin ƙamus na mahallinsa. Wannan tsarin yana taimakawa tantance ko ya kamata a ƙara voxel zuwa ƙamus na mahallin ba tare da duban matrix mai wahala ba, yana haifar da raguwa mai mahimmanci a cikin rikitarwar lissafi da lokacin aiki.
3. Lissafin yiwuwar
   Don ƙididdige yuwuwar entropy, duka tsayin jeri da mahallin voxels ɗin sa dole ne a yi la'akari da su. A cikin wannan tsarin, mun ƙirƙiri mai rikodin daidaitawa wanda zai fara ƙididdige manyan iyakoki na sama da ƙasa na yuwuwar ga kowane rukuni daga ƙamus na mahallin, sannan mu ɓoye shi daga baya. Da farko, muna gina bishiyar binaryar bisa tsarin sarkar Markov. Ta hanyar tsallake zama na voxels, muna ba da ƙima na 1 da 0 zuwa shagaltar da voxels, bi da bi, da lissafin yuwuwar bisa tsarin bishiyar. Fara daga tushen tushen, lokacin da aka shagaltar da voxel, muna yin rikodin kumburin yaro na hagu kamar 1. In ba haka ba, muna yiwa alamar yaron dama alama kamar 0 kuma ci gaba zuwa mataki na gaba na hukunci da rarraba. Za'a iya samun dabarar lissafi don yuwuwar gudu na voxels da aka mamaye a cikin Eq. (4).

Don tsayin gudu ƙasa da ko daidai da n, ana iya samun 2n na nodes na bishiyar da ke wakiltar jihohin zama na voxels. Don haka, yuwuwar kowane voxel da aka mamaye yana wakiltar yiwuwar haɗin gwiwa mai zaman kansa na ratsa duk jihohin da ke farawa daga tushe kuma suna ƙarewa a kowane kumburin bishiyar mara haihuwa. Dangane da Eq. (4), don yin rikodin lissafin lissafi akan zama na jerin voxel, muna buƙatar yuwuwar babba da ƙasa na jeri, kamar yadda aka nuna a Eq. (5).

Yin amfani da wannan hanyar, za mu iya amfani da kaddarorin daidaitawa na ƙididdige ƙididdiga don daidaita ƙimar ƙimar yiwuwar kowace alama bisa ingantacciyar ƙirar ƙima da mitar kowace alama a cikin jerin alamun yanzu. Wannan yana ba mu damar ƙididdige manyan iyakoki na sama da ƙasa na yuwuwar tarawa na voxels da aka shagaltar da su kuma mu kammala aikin ɓoyewa.

Gwaji

Cikakken Bayani

1. Saitin bayanai. Don tabbatar da aikin hanyar da aka tsara, an gudanar da gwaje-gwaje masu yawa sama da maki 16 na bayanan girgije waɗanda za a iya saukewa daga Ref. [40], kamar yadda aka nuna a hoto na 3, wanda Fig. 3 (a) - 3 (l) hotuna ne masu yawa, da Figs. 3 (m) - 3 (p) gine-gine ne tare da ƙananan maki. Figs. 3(a) - 3(h) jerin bayanan gajimare ne da aka yi amfani da su na babban juzu'i na shawarwarin sararin samaniya guda biyu da aka samu daga Microsoft. Figs. 3 (i) - 3 (l) an zaɓi su daga 8i voxelized cikakken jikin bayanan bayanan bayanan girgije. Ragowar gajimare masu girman gaske a cikin figs. 3 (k) - 3 (p) facade na tsaye da bayanan gine-gine.
2. Ma'aunin kimantawa. Ana kimanta aikin hanyar da aka tsara a cikin sharuddan bit a kowane aya (BPP). BPP tana nufin jimlar raƙuman raƙuman ruwa da ke tattare da bayanan haɗin gwiwar da ke haɗe zuwa wurin. Ƙananan ƙimar, mafi kyawun aikin. inda Sizedig ke wakiltar adadin ragowar da ke tattare da daidaita bayanan bayanan girgije, kuma k yana nufin adadin maki a cikin gajimaren batu na asali.
3. Alamu. Mun fi kwatanta hanyarmu da wasu algorithms na asali, gami da: PCL-PCC: matsawa na tushen octree a cikin PCL; G-PCC (samfurin gwajin intra-coders MPEG) da kuma interEM (samfurin gwaji na inter-coders na MPEG) sun yi niyya ga matsi guda-frame da matsa lamba mai yawa bi da bi; Silhouette 3D (S3D) [41] da Silhouette 4D (S4D) [42] manufa guda firam da matsa lamba mai yawa-firam, bi da bi.
   Don PCL, muna amfani da tsarin matsawa gajimare na octree a cikin PCL-v1.8.1 don matsawar lissafi kawai. Mun saita sigogin ƙuduri na octree daga madaidaicin maki da ƙudurin voxel. Don G-PCC (TM13-v11.0), mun zaɓi nau'in lissafi mara asara - yanayin halayen rashin hasara a cikin yanayin tsinkayar octree, barin sigogi azaman tsoho. Don interEM (tmc3v3.0), muna amfani da sakamakon gwaji a ƙarƙashin juzu'i marasa asara da halaye marasa asara azaman kwatance[43]. Za S3D
   da S4D, muna bin yanayin tsoho da sigogi.
4. Hardware. Ana aiwatar da algorithm da aka tsara a cikin Matlab da C++ ta amfani da wasu ayyuka na PCL-v1.8.1. An gwada duk gwaje-gwaje akan kwamfutar tafi-da-gidanka tare da Intel Core i7- 8750 CPU @2.20 GHz tare da ƙwaƙwalwar 8 GB.

Sakamako na Matsa Matsakaicin-Frame Point Cloud

1. Sakamakon matsi na hotuna na jerin bayanan girgije masu yawa
   Tebur 1 yana nuna aikin mu na mahallin mahallin madaidaicin rashi maras ma'ana gajimare matsawa algorithms idan aka kwatanta da PCL-PCC, G-PCC da hanyoyin S3D akan hotunan jerin bayanan girgije masu yawa. Ana iya gani daga Tebu 1 cewa ga duk ma'anar girgije na jeri iri ɗaya, hanyar da aka tsara ta cimma mafi ƙarancin matsawa BPP idan aka kwatanta da sauran hanyoyin. Algorithm ɗinmu yana ba da matsakaicin riba daga -1.56% zuwa -0.02% akan S3D, kuma yana samun riba daga -10.62% zuwa -1.45% akan G-PCC. Yana nuna advan mafi bayyanetage, wato, nasarorin aikin matsawa na ƙirar algorithm da aka tsara daga -10.62% zuwa - 1.45%; Don PCL-PCC, algorithm da aka tsara yana nuna riba kusan ninki biyu akan duk jeri, kama daga -154.43% zuwa -85.39%.
2. Sakamakon matsi na bayanan gajimare masu girman gaske
   Saboda S3D ba zai iya aiki a cikin wannan yanayin ba, kawai muna kwatanta yanayin mahallin mu ne kawai mai jagorar hasarar ma'auni na matsawar gajimare tare da PCL-PCC da hanyoyin G-PCC akan manyan bayanan girgije mara iyaka. Bugu da ƙari, algorithm ɗin mu yana samun babban aiki tare da G-PCC da PCL-PCC, kamar yadda aka nuna a cikin Table 1. Sakamako ya nuna cewa matsakaicin ribar BPP daga - 8.84% zuwa -4.35% an kama idan aka kwatanta da G-PCC. Don PCL-PCC, algorithm ɗinmu da aka tsara yana nuna ƙarin fayyace advantages, tare da ribar da ke jere daga -34.69% zuwa -23.94%.
3. Takaitawa
   Don samar da ƙarin mahimmin kwatancen sakamakon matsewar gajimare na firam guda ɗaya, Tebura 2 yana gabatar da matsakaicin sakamako tsakanin hanyar matsi mai jagorar mahallin sararin samaniya da sauran hanyoyin ma'auni na zamani. Idan aka kwatanta da S3D, hanyar da muka gabatar tana nuna matsakaicin ribar da aka samu daga - 0.58% zuwa - 3.43%. Dangane da G-PCC da PCL-PCC, matsakaicin ribar da aka samu ya kai aƙalla -3.43% da -95.03% bi da bi. Binciken gwaji yana bayyana cewa hanyar matsi na mahallin mahallin mu ya zarce S3D na yanzu, G-PCC da PCL-PCC ta wani babban gefe. Don haka, zai iya gamsar da buƙatun matsawa maras nauyi na ma'aunin lissafi na girgije don nau'ikan yanayi daban-daban, misali, rarrabawa mai yawa ko kaɗan, kuma tasirin hanyarmu ta kasance koyaushe.
4. Sakamako na Multi-frame Point Cloud Compression
   Muna ƙididdige abubuwan da muka gabatar na sararin samaniya-lokacin mahallin mahallin madaidaicin madaidaicin juzu'in juzu'i na matsawa algorithm akan abubuwan matsawa na yanzu kamar S4D, PCL-PCC, G-PCC da interEM. Hotunan jerin bayanan girgije masu yawa ne kawai ake amfani da su a wannan gwaji. An kwatanta sakamakon a cikin.

Tebur 1. Kwatancen BPP na algorithm na matsawa mai jagorar sararin samaniya da hanyoyin tushe.

Tebur 2. Kwatanta BPP tare da algorithms na zamani akan bayanan girgije mai ma'ana guda ɗaya.

Tebur 3. Kamar yadda muke iya gani, bayan ingantawa a cikin yanayin tsinkaya da rikodin ƙididdiga, algorithm da aka tsara yana nuna fifiko akan duk jerin gwaji. Musamman, idan aka kwatanta da interEM da G-PCC, algorithm da aka ba da shawarar yana nuna manyan nasarori daga -51.94% zuwa -17.13% da -46.62% zuwa -5.7%, bi da bi. Idan aka kwatanta da S4D, algorithm ɗin da aka tsara yana nuna ingantaccen haɓakawa daga -12.18% zuwa -0.33%. Dangane da PCL-PCC, algorithm ɗinmu da aka tsara ya kusan raguwa sama da duk jerin gwajin.
Bugu da ƙari kuma, muna taƙaita sakamakon matsawa da ribar hanyar da aka tsara akan jerin bayanan girgije mai girma na hoto, wanda aka jera a cikin Teburin 4. A matsakaita, yana ba da riba tsakanin -11.5% da -2.59% idan aka kwatanta da sararin mahallin mahallin jagorar girgije. algorithm matsawa na geometry da aka gabatar a baya. Bugu da ƙari, yana nuna matsakaicin matsakaicin ribar -19% idan aka kwatanta da G-PCC kuma ya sami matsakaicin ribar coding na -24.55% idan aka kwatanta da interEM. Bugu da ƙari, idan aka kwatanta da S3D da S4D, yana samun fiye da -6.11% da -3.64% akan matsakaita. Binciken gwajin gabaɗaya ya nuna cewa hanyar matsawa gajimare mai jagorar mahallin mahallin sararin samaniya na iya yin cikakken amfani da yanayin sararin samaniya da na ɗan lokaci na yadudduka da ke kusa a cikin firam ɗin intra-frames da tsaka-tsakin firam. Hakanan muna haɓaka zaɓin mahallin duniya da ƙirar yuwuwar mai rikodin lissafin don samun ƙaramin ɗan ƙaramin kuɗi. Hanyar da aka tsara ta zarce aikin algorithms na zamani, ta yadda za a iya biyan buƙatun matsewar gizagizai na gizagizai marasa asara a cikin yanayin aikace-aikacen multimedia kamar hotuna masu ƙarfi.

Tebur 3. Kwatancen Bit-per-point na mu na sarari-lokaci mahallin jagorar matsawa algorithm da hanyoyin tushe.

Tebur 4. Kwatancen Bit-per-point tare da algorithms na zamani akan bayanan girgije mai ma'ana da yawa.

Nazarin Ablation
Muna yin nazarin ɓarna a kan ɓoyayyiyar tsinkaya sama da 8i voxelized cikakkun bayanan bayanan girgije don nuna tasirin ɓangaren. Ana iya gani daga Teburin 5 cewa haɓakawa yana nuna ingantaccen riba na -70% akan matsawar girgije mai ma'ana da yawa da kuma - 60% akan matsawar girgije mai ma'ana guda ɗaya akan ƙididdige ƙididdiga marasa ɓarna.
Na gaba, za mu yi gwajin ɓarna akan ƙididdige ƙididdiga don nuna ingancin ƙamus na mahallin. Kamar yadda aka nuna a cikin Tebu 6, ingantaccen haɓakawa na -33% akan matsawar gajimare mai nau'i-nau'i da yawa da na -41% akan matsawar girgije mai ma'ana guda ɗaya akan ƙididdige ƙididdiga ba tare da ƙamus na mahallin a cikin hanyarmu ba.

Amfanin Lokaci
Muna gwada amfani da lokaci don kimanta hadaddun algorithm kuma kwatanta hanyoyin da aka tsara tare da wasu. Ana nazarin hadadden algorithm ta encoders da decoders da kansa, wanda aka jera a cikin Table 7. Kamar yadda muke iya gani, G-PCC, interEM da PCL-PCC na iya cimma lokacin ɓoyewa na ƙasa da 10 s da lokacin yankewa na ƙasa da 5 s don bayanan girgije mai girman hoto. Hakanan suna yin aiki da kyau a cikin manyan bayanan girgije mara iyaka idan aka kwatanta da wasu. Algorithms ɗin da muke samarwa suna ɗaukar kusan 60 s da 15 s don ɓoyewa da yanke jerin jerin hotuna, har ma akan facade da bayanan gajimare na gine-gine. Akwai ciniki tsakanin bitrates da saurin matsawa. Idan aka kwatanta da S3D da S4D, waɗanda ke ɗaukar ɗarurruwan daƙiƙa don ɓoyewa, hanyar cin lokaci namu na iya nuna fifiko.
A taƙaice, amfani da lokaci na hanyoyin da aka tsara yana da matsakaici tsakanin duk kwatankwacin algorithms amma har yanzu ya zama dole don haɓakawa.

Ƙarshe

A cikin wannan takarda, muna ba da shawarar hanyar da za ta jagoranci mahallin yanayi don rashin hasarar ma'ana ga ma'aunin gizagizai. Muna la'akari da yanki mai kauri mai kauri a matsayin naúrar shigarwa kuma muna ɗaukar yanayin ƙididdiga na lissafi dangane da algorithm mai siyar da balaguro, wanda ya shafi duka intra-frame da inter-frame. Haka kuma, muna yin cikakken amfani da bayanan mahallin duniya da mai rikodin lissafin daidaitawa dangane da sabuntawa cikin sauri-sauri don cimma matsi mara asara da sakamakon yankewar gizagizai. Sakamakon gwaji yana nuna tasirin hanyoyinmu da fifikon su akan karatun da suka gabata. Don aiki na gaba, muna shirin ƙara nazarin gabaɗayan rikitarwa na algorithm, ta hanyar rage ƙayyadaddun algorithm don cimma ƙimar matsawa mai sauri da ƙananan sakamakon matsawa. Ƙarƙashin ƙimar bit da hanyar tallafi na ainihin lokaci/ƙananan jinkiri ana so sosai a cikin nau'ikan fage daban-daban.

Magana

1. MI XX, YANG BS, Dong Z, et al. Cire kan iyaka na 3D mai sarrafa kansa da haɓakawa ta amfani da gajimare na MLS [J]. IEEE ma'amaloli akan tsarin sufuri na hankali, 2022, 23 (6): 5287 - 5297. DOI: 10.1109 / TITS.2021.3052882
2. Dong Z, LIANG FX, YANG BS, et al. Rijistar manyan gizagizai na Laser na'urar daukar hotan takardu: review da benchmark [J]. Mujallar ISPRS na daukar hoto da hangen nesa nesa, 2020, 163: 327-342. DOI: 10.1016/j.isprsjprs.2020.03.013
3. GRAZIOSI D, NAKAGAMI O, KUMA S, et al. An wuceview Ayyukan daidaita ma'aunin matsi mai gudana: tushen bidiyo (V-PCC) da tushen geometry (G-PCC) [J]. Ma'amalar APSIPA akan sigina da sarrafa bayanai, 2020, 9: e13
4. DE QUEIROZ RL, CHOU P A. Matsi na 3D batu gajimare ta amfani da wani yanki-adaptive matsayi canji [J]. IEEE ma'amaloli akan sarrafa hoto, 2016, 25(8): 3947-3956. DOI: 10.1109/TIP.2016.2575005
5. BLETTERER A, PAYAN F, ANTONINI M, et al. Nuna matsawar girgije ta amfani da taswirorin zurfin [J]. Hoton lantarki, 2016, 2016 (21): 1-6
6. MEKURIA R, BLOM K, CESAR P. Zane, aiwatarwa, da kimantawa na codec girgije mai ma'ana don bidiyo na immersive [J]. IEEE ma'amaloli akan ma'amaloli da tsarin don fasahar bidiyo, 2017, 27 (4): 828 - 842. DOI: 10.1109/ TCSVT.2016.2543039
7. DE QUEIROZ RL, CHOU P A. Motion-compensated compressing voxelized point Clouds [J]. IEEE ma'amaloli akan sarrafa hoto, 2017, 26 (8): 3886-3895. DOI: 10.1109/TIP.2017.2707807
8. CAO C, PREDA M, ZAHARIA T. 3D batu na girgije: bincike [C]// Taron kasa da kasa na 24th akan 3D Web Fasaha. ACM, 2019: 1–9. DOI: 10.1145/3329714.3338130
9. GRAZIOSI D, NAKAGAMI O, KUMA S, et al. An wuceview Ayyukan daidaita ma'aunin matsi mai gudana: tushen bidiyo (V-PCC) da tushen geometry (G-PCC) [J]. Ma'amalar APSIPA akan sigina da sarrafa bayanai, 2020, 9(1): e13. DOI: 10.1017/atsip.2020.12
10. HUANG Y, PENG JL, KUO CJ, et al. Ƙididdigar juzu'i na ci gaba na tushen Octree na girgije mai ma'ana [C]//Taro na 3 na Yurographics/IEEE VGTC akan Zane-zane na Tushen. IEEE, 2016: 103-110
11. FAN YX, HUANG Y, PENG J L. Matsawar gajimare bisa tushen gungu na matsayi na matsayi [C]//Siginar Asiya-Pacific da Ƙungiyar Gudanar da Bayani na Shekara-shekara da Taro. IEEE, 2014: 1 - 7. DOI: 10.1109/APSIPA.2013.6694334
12. DRICOT A, ASCENSO J. Miyan alwatika mai ɗabi'a mai daidaitawa don tushen ma'anar gajimare coding [C]//Bita na 21st International Workshop on Multimedia Signal Processing (MMSP). IEEE, 2019: 1 - 6. DOI: 10.1109/ MMSP.2019.8901791
13. HE C, RAN LQ, WANG L, et al. Matsa lamba saitin saman da ya dogara akan sifar ƙirar ƙira [J]. Multimedia kayan aikin da aikace-aikace, 2017, 76 (20): 20545-20565. DOI: 10.1007/s11042-016-3991-0
14. IMDAD U, ASIF M, AHMAD M, et al. Matsa lamba mai girma uku da matsawa gajimare ta amfani da nau'ikan nau'ikan digiri na ɗaya [J]. Misali, 2019, 11 (2): 209. DOI: 10.3390/sym11020209
15. SUN XB, MA H, SUN YX, et al. Algorithm na matsawa gajimare na sabon labari dangane da tari [J]. IEEE mutum-mutumi da haruffan aiki da kai, 2019, 4(2): 2132-2139. DOI: 10.1109/LRA.2019.2900747
16. DE OLIVEIRA RENTE P, BRITES C, ASCENSO J, et al. Girgizar gizagizai na tushen faifan ma'ana 3D mai ƙima [J]. Ma'amaloli na IEEE akan multimedia, 2019, 21(2): 284-299. DOI: 10.1109/TMM.2018.2859591
17. ISO. Matsawar girgije mai tushen geometry (G-PCC): ISO/IEC 23090-9 [S]. 2021
18. DRICOT A, ASCENSO J. Hybrid octree-plane point cloud geometry codeing [C]//Taron Gudanar da Siginar Turai na 27 (EUSIPCO). IEEE, 2019: 1-5
19. ZHANG X, GAO W, LIU S. Bangaren juzu'i na zahiri don matsawar gajimare ma'ana [C]// Gabatar da taron 2020 Data Compression (DCC). IEEE, 2020: 73-82. DOI: 10.1109/DCC47342.2020.00015
20. QUACH M, VALENZISE G, DUFAUX F. Koyan juzu'i na juyi don matsawa ma'anar gizagizai na gajimare [C]// Taron kasa da kasa na IEEE akan Tsarin Hoto (ICIP) na 2019. IEEE, 2019: 4320- 4324. DOI: 10.1109/ICIP.2019.8803413
21. HUANG TX, LIU Y. 3D batu na lissafin lissafi na girgije akan zurfin koyo [C]// Taron kasa da kasa na ACM na 27 akan Multimedia. ACM, 2019: 890-898. DOI: 10.1145/3343031.3351061
22. GUARDA AFR, RODRIGUES NMM, PEREIRA F. Nuna lambar gajimare: Ƙarfafa tsarin ilmantarwa mai zurfi [C]//Taro na Hotuna (PCS). IEEE, 2020: 1-5. DOI: 10.1109/PCS48520.2019.8954537
23. WANG JQ, ZHU H, MA Z, et al. Koyon ma'aunin gizagizai na geometry [EB/OL]. [2023-09-01]. https://arxiv.org/abs/1909.12037.pdf
24. AINALA K, MEKURIA RN, KHATHARIYA B, et al. Ingantattun Layer ment don matsewar gajimare mai tushen octree tare da tsinkayar ap⁃ kusancin jirgin sama [C]// SPIE Optical Engineering+Aikace-aikace. SPIE, 2016: 223- 231. DOI: 10.1117/12.2237753
25. SCHWARZ S, HANNUKSELA MM, FAKOUR-SEVOM V, et al. 2D video codeing na volumetric video data [C]//Hoto Codeing Symposium (PCS). IEEE, 2018: 61-65. DOI: 10.1109/PCS.2018.8456265
26. FAKOUR SEVOM V, SCHWARZ S, GABBOUJ M. Geometry-Jagorar bayanan bayanan 3D don tsinkayar tushen tsinkayar gajimare mai ƙarfi [C]//Bita na 7 na Turai akan Tsarin Bayanan Kayayyakin gani (EUVIP). IEEE, 2019: 1-6. DOI: 0.1109/EUVIP.2018.8611760
27. KATHARIYA B, LI L, LI Z, et al. Rashin hasarar madaidaicin ma'aunin lissafi na gajimare tare da diyya ta tsaka-tsaki da tsinkayar mai siyar da balaguro [C]// Taron Matsalolin Bayanai. IEEE, 2018: 414. DOI: 10.1109/ DCC.2018.00067
28. ISO. Kayayyakin tushen coding na tushen bidiyo (V3C) da matsawar girgije mai tushen bidiyo: ISO/IEC 23090-5 [S]. 2021
29. PARK J, LEE J, PARK S, et al. Taswirar taswirar zama na tushen tsinkaya don matsawar girgije mai maki 3D [J]. Ma'amaloli na IEIE akan sarrafa wayo & sarrafa kwamfuta, 2020, 9(4): 293-297. DOI: 10.5573/ieespc.2020.9.4.293
30. Costa A, DRICOT A, BRITES C, et al. Ingantattun faci don ma'aunin MPEG V-PCC [C]//IEEE 21st International Workshop on Multimedia Signal Processing (MMSP). IEEE, 2019: 1 - 6. DOI: 10.1109/ MMSP.2019.8901690
31. KAMMERL J, BLODOW N, RUSU RB, et al. Matsi na ainihin lokacin rafukan gajimare [C]//Tsarin 2012 IEEE taron kasa da kasa kan Robotics da Automation. IEEE, 2012: 778 - 785. DOI: 10.1109 / ICRA.2012.6224647
32. PCL. Nuna ɗakin karatu na girgije. [EB/OL]. [2023-09-01]. http://pointclouds.org/
33. THANOU D, CHOU PA, FROSSARD P. Tsarin tushen zane na jerin gajimare mai ƙarfi na 3D [J]. IEEE ma'amaloli akan sarrafa hoto, 2016, 25(4): 1765-1778. DOI: 10.1109/TIP.2016.2529506
34. LI L, LI Z, ZAKHARCHENKO V, et al. Hasashen motsi na 3D na ci gaba don matsewar halayen girgije na tushen tushen bidiyo [C]//Taron matsawa bayanai (DCC). IEEE, 2019: 498-507. DOI: 10.1109/DCC.2019.00058
    ZHAO LL, MA KK, LIN XH, et al. Matsawar girgije na LiDAR na ainihin lokaci ta amfani da tsinkayar Bi-directional da kewayon-daidaita-take-take-kodi [J]. IEEE ma'amaloli akan watsa shirye-shirye, 2022, 68 (3): 620 - 635. DOI: 10.1109/TBC.2022.3162406
35. LIN JP, LIU D, LI HQ, et al. M-LVC: Hasashen firam ɗin da yawa don koyi matsawar bidiyo [C]// IEEE/CVF taron akan hangen nesa na Kwamfuta da Gane Tsarin. IEEE, 2020: 3543 - 3551. DOI: 10.1109/ CVPR42600.2020.00360
36. YANG R, MENTZER F, VAN GOOL L, et al. Koyo don matsawa bidiyo tare da inganci na matsayi da haɓakawa akai-akai [C]//IEEE/CVF taron akan hangen nesa na Kwamfuta da Gane Ƙa'idar. IEEE, 2020: 6627-6636.DOI: 10.1109/CVPR42600.2020.00666
37. KAYA EC, TABUS I. Rasa matsi na ma'ana jerin girgije ta amfani da jerin ingantattun samfuran CNN [J]. Samun damar IEEE, 2022, 10: 83678-83691. DOI: 10.1109/ACCESS.2022.3197295
38. DING S, MANNAN MA, POO A N. Akwatin ɗaure mai daidaitacce da gano tsangwama na tushen octree a cikin 5-axis machining na filayen kyauta [J]. Zane-zanen kwamfuta, 2004, 36 (13): 1281-1294
39. ALEXIOU E, VIOLA I, BORGES TM, et al. Cikakken bincike na aikin karkatar da ƙima a cikin matsawar girgije ta MPEG [J]. Ma'amalar APSIPA akan sigina da sarrafa bayanai, 2019, 8: e27. doi:10.1017/ ATSIP.2019.20
40. PEIXOTO E. Intra-frame compression of point cloud geometry ta amfani da bazuwar dyadic [J]. Wasiƙun Gudanar da Siginar IEEE, 2020, 27: 246-250. DOI: 10.1109/LSP.2020.2965322
41. RAMALHO E, PEIXOTO E, MEDEIROS E. Silhouette 4D tare da zaɓin mahallin: rashin hasarar joometry na girgije mai ƙarfi [J]. IEEE Haruffa Tsara Siginar, 2021, 28: 1660 - 1664. DOI: 10.1109/ lsp.2021.3102525
42. ISO. Sharuɗɗan gwaji gama gari don takaddar G-PCC N00106: ISO/IEC JTC 1/ SC 29/WG 7 MPEG [S]. 2021

Tarihin rayuwa

• ZHANG Huiran ta sami digiri na BE da ME a Makarantar Geodesy da Geomatics da Jiha Key Laboratory of Information Engineering in Surveying Mapping and Remote Sensing, duka daga Jami'ar Wuhan, China a 2020 da 2023, bi da bi. A halin yanzu ita ce ma'aikaciyar bincike ta Cibiyar Nazarin Tsare-Tsare da Tsare Tsare Tsaren Birane ta Guangzhou, na kasar Sin. Abubuwan bincikenta sun haɗa da sarrafa bayanan girgije da matsawa. Ta shiga cikin ayyuka da yawa da suka shafi fannin hangen nesa kuma ta buga takarda ɗaya a cikin Geomatics da Kimiyyar Bayani na Jami'ar Wuhan.
• DONG Zan (dongzhenwhu@whu.edu.cn) ya sami digirin sa na BE da PhD a fannin sanin nesa da daukar hoto daga jami'ar Wuhan ta kasar Sin a shekarar 2011 da 2018, bi da bi. Shi farfesa ne tare da Maɓallin Maɓalli na Jiha na Injiniyan Watsa Labarai a cikin Bincike, Taswirar Taswira da Nesa (LIESMARS), Jami'ar Wuhan. Bukatun bincikensa sun haɗa da sake gina 3D, fahimtar yanayi, sarrafa girgije da kuma aikace-aikacen su a cikin tsarin sufuri mai hankali, biranen tagwayen dijital, ci gaba mai dorewa na birane da robotics. Ya samu karramawa sama da 10 daga gasa daban-daban na kasa da kasa da kuma buga takardu kusan 60 a mujallu da tarurruka daban-daban.
  WANG Mingsheng Ya sami digirin sa na BE a Kwalejin Kimiyya da Fasaha ta Kwamfuta daga Jami'ar Jilin ta kasar Sin a shekarar 2001, sannan ya sami digiri na ME a Makarantar Kimiyyar Kwamfuta da Injiniya daga Jami'ar Fasaha ta Kudancin kasar Sin da ke kasar Sin a shekarar 2004. A halin yanzu babban injiniya ne a fannin tsara biranen Guangzhou. & Cibiyar Nazarin Binciken Zane, China. Abubuwan bincikensa sun haɗa da aikace-aikacen kwamfuta da software, physiography, da binciken. Ya samu karramawa sama da 20 daga gasa daban-daban na kasa da kuma buga takardu kusan 50 a cikin mujallu da tarurruka daban-daban.

DOI: 10.12142/ZTECOM.202304003
https://kns.cnki.net/kcms/detail/34.1294.TN.20231108.1004.002.html, wanda aka buga akan layi Nuwamba 8, 2023
An karɓi rubutun: 2023-09-11

Takardu / Albarkatu

Algorithm na Jagorar ZTE don Matsa Matsalolin Geometry Cloud maras Rasa [pdf] Jagorar mai amfani Algorithm Jagoran don Matsin Geometry Cloud maras Rasa, Jagorar, Algorithm don Matsa Matsalolin Geometry mara Rasa, Matsa Matsalolin Geometry Cloud, Matsa Matsarin Geometry, Matsin Geometry

Magana

• Manual mai amfani