Amamodeli avamile amayunithi okucubungula isignali yekhamera yokucabanga eshisayo

Iyunithi yokucubungula isignali yekhamera yokucabanga eshisayo iwukhiye wesistimu yokucabanga eshisayo, enesibopho sokuguqula amasignali aluhlaza aqoqwe yi-infrared. inzwa kumasignali wekhwalithi ephezulu, asebenzisekayo. Nawa amamodeli ajwayelekile futhi asebenza kahle kakhulu.

Amaphrosesa wesignali yedijithali yochungechunge lwe-TMS320 asebenzisa ukwakheka kwegama lomyalelo omude kakhulu (VLIW), wokwenza imisebenzi eminingi ngokufana ngaphakathi komjikelezo wewashi elilodwa, namaza ewashi afinyelela ku-gigahertz, okuwenza asebenze kahle kakhulu ekucubunguleni idatha yokucabanga eshisayo. Iyunithi edidiyelwe yokucubungula ezintantayo ingasebenzisa ngempumelelo ama-algorithms ayinkimbinkimbi, njenge-Fourier transform esheshayo (FFT), yokuhlaziya imvamisa yesignali nomsindo wokuhlunga. Lolu chungechunge luphinde lusekele ukutholwa kwedatha yeziteshi eziningi okuvumela izinzwa eziningi ze-infrared ukuthi zixhunywe ngesikhathi esisodwa, ukuhlangabezana nokulungiswa kwesithombe esishisayo esihlukile kanye nezidingo zezinga lozimele, futhi lisetshenziswa kabanzi ekuhlolweni kwezimboni nokuqapha ukuphepha.

Uchungechunge lwe-Jetson ludizayinelwe ukusebenzisa ikhompyutha enqenqemeni kanye nezinhlelo zokusebenza zobuhlakani bokwenziwa. Ukuthatha i-Jetson Xavier NX njengesibonelo, ihlanganisa amakhompiyutha asebenza kahle kakhulu njenge-GPU nama-accelerator okufunda okujulile. Ikhono lekhompuyutha elihambisanayo le-GPU lingaqedela ngokushesha ukucutshungulwa kwesignali eyisisekelo, futhi amandla ekhompuyutha ye-AI enamandla asekela ukusebenzisa ama-algorithms enethiwekhi ye-neural afana namanethiwekhi e-convolutional neural (CNN). Ekuqapheni umlilo wehlathi, ingakwazi ukukhomba ngokuzenzakalelayo imithombo yomlilo kanye nentuthu; ezinhlelweni ezisizayo zokushayela izithombe ezishisayo, kungavimbela ukutholwa nokugwema. Ngaphezu kwalokho, lolu chungechunge lunokuxhumana okucebile nokunwebeka okuqinile, okwenza kube lula ukuhlanganisa namanye amadivaysi.

I-Hikvision inamandla aqinile emkhakheni wokubhekwa kwevidiyo. Iyunithi yayo yokucubungula isignali yekhamera yesithombe esishisayo isebenzisa i-chip yokucubungula izithombe ezithuthukisiwe esebenza kahle kakhulu. Ngobuchwepheshe obuhambisanayo obunemicu eminingi, ingakwazi ukucubungula ngokushesha amasiginali avela eziteshini zokucabanga ezishisayo eziningi ngesikhathi esisodwa, iqinisekise ukuthi izithombe ze-imaging ezishisayo ezicacile nezizinzile zingaphuma ngisho nasezigcawini eziyinkimbinkimbi. Mayelana nokuhlaziya, ihlanganisa ama-algorithms okufunda ajulile akwazi ukuhlonza okuqondiwe futhi aqaphe ukuziphatha okungajwayelekile. Ngaphezu kwalokho, iyunithi yokucubungula isignali inokuhambisana okuhle futhi ingaxhunywa ngaphandle komthungo ngezinzwa ezihlukahlukene ze-infrared kanye nemishini yokuphela, futhi isetshenziswa kabanzi ekuvikelekeni kwedolobha, ukuqapha kwezezimali kanye nezinye izinkambu.

I-RK399Pro iyisistimu-on-chip (i-SoC) esebenza kahle kakhulu esebenzisa i-architecture ye-processor eyisisekelo eyisithupha, elinganisa amandla ekhompuyutha anamandla nokusetshenziswa kwamandla okuphansi. Injini yekhodekhi yehadiwe ye-Hardware isekela amafomethi ombhalo wekhodi wevidiyo njenge-H.264 ne-H.265, ikwenze kube lula ukugcina nokudlulisa amasiginali wokuthwebula izithombe ezishisayo. ukuxhumana okucebile kwenza kube lula ukuxhuma izinzwa ezihlukahlukene ze-infrared namadivayisi angaphandle, futhi idlala indima ebalulekile ekuvikelekeni kwasekhaya okuhlakaniphile, ukuqapha imishini emincane yezimboni nezinye izimo.

amamodeli avamile, anezinzuzo zawo zezobuchwepheshe, akhuthaza ukusetshenziswa okusabalele kobuchwepheshe bezithombe ezishisayo emikhakheni ehlukene. Ngokuzayo, ngokuthuthuka kobuchwepheshe, ukusebenza okwengeziwe, amandla aphansi kanye namamodeli eyunithi yokucubungula isignali enothile ngomsebenzi azovela, okuthuthukisa ukusebenza kanye nenani lesicelo lamakhamera wokuthwebula izithombe ezishisayo.