Hanyar Kwaikwayi da Bincike Mai Zurfi na Tasirin Gani a cikin Na'urori na Auna Lokacin Tafiya (ToF) na Kamara

1. Gabatarwa

Na'urori na Auna Lokacin Tafiya (ToF) na Kamara suna ba da hanya mai sauri da dacewa don samun bayanan muhalli na 3D ta hanyar auna lokacin tafiya da dawowa na hasken da aka fitar da shi da gangan. Wannan takarda tana gabatar da cikakkiyar hanyar kwaikwayi don kimanta aikin na'urar, daidaito, da kuma fahimtar tasirin da aka lura da shi ta hanyar gwaji, tare da mai da hankali kan cikakken kwaikwayin siginar gani.

2. Ka'idojin Auna Lokacin Tafiya (ToF)

Na'urorin ToF suna lissafin nisa a kowane pixel ta hanyar auna lokacin da haske ke ɗauka daga tushe zuwa abu sannan ya koma na'urar gano shi (detector).

2.1 Auna Lokacin Tafiya Kai Tsaye (D-ToF)

Yana auna lokacin tafiya da dawowa kai tsaye ta amfani da gajerun bugun jini (nanosecond). Duk da cewa a zahiri yana da sauƙi, yana fama da ƙarancin rabo na sigina zuwa hayaniya (SNR) saboda buƙatar na'urorin lantarki masu sauri (GHz), kamar yadda Jarabo da sauransu (2017) suka lura. Ana lissafin nisa $d$ a sauƙaƙe kamar haka: $d = \frac{c \cdot \Delta t}{2}$, inda $c$ shine saurin haske kuma $\Delta t$ shine lokacin da aka auna.

2.2 Auna Lokacin Tafiya na Tushen Haɗin Kai (C-ToF/P-ToF)

Hanyar da ta fi yawa a cikin na'urorin kasuwanci. Tana amfani da hasken igiyar igiyar ruwa mai ci gaba da daidaitawa (AMCW). Ana auna canjin lokaci (phase shift) $\phi$ tsakanin siginonin da aka fitar da waɗanda aka karɓa, kuma ana samun zurfi daga gare shi: $d = \frac{c \cdot \phi}{4\pi f_{mod}}$, inda $f_{mod}$ shine mitar daidaitawa (yawanci a cikin MHz). Ana aiwatar da wannan ta amfani da Na'urorin Haɗa Hotuna (PMD) a kowane pixel da kuma soke daidaitawa (Lock-In demodulation) (Schwarte da sauransu, 1997; Lange, 2000).

Bayanin Hoto na 1: Zane na na'urar ToF na kamara ta amfani da dabarar AMCW. Tsarin ya ƙunshi tushen haske mai daidaitawa (LED/VCSEL), ruwan tabarau, matrix na pixel tare da haɗaɗɗun da'irori na soke daidaitawa (PMD), na'urar canza sigina zuwa lambobi (A/D converter), mai sarrafa jerin gwano, da mai sarrafa mai gida don lissafin taswirar zurfi.

3. Hanyar Kwaikwayi da Ake Shawarwari

Babban gudunmawar shi ne tsarin kwaikwayi na tushen binciken haske (raytracing) wanda ke amfani da tsawon hanyar gani a matsayin babban ma'auni don lissafin zurfi, wanda ya wuce ƙananan samfuran daga maki zuwa maki.

3.1 Hanyar Tsawon Hanyar Gani ta Tushen Binciken Haske (Raytracing)

Maimakon kwaikwayin hanyoyin nunawa kai tsaye kawai, hanyar tana bin hasken ta hanyoyin gani masu sarkakiya. Jimillar tsawon hanyar gani (OPL) don haske ana bayar da shi ta $OPL = \int_{}^{} n(s) \, ds$, inda $n$ shine ma'aunin karkatar haske (refractive index) tare da hanyar $s$. Wannan OPL yana da alaƙa kai tsaye da canjin lokaci (phase shift) da aka auna a cikin tsarin C-ToF.

3.2 Aiwarta a cikin Zemax OpticStudio da Python

Ana yin binciken haske (ray tracing) a cikin Zemax OpticStudio don ƙirar ruwan tabarau, tushe, da hulɗar abubuwa tare da ingantaccen aminci. Na'urar baya ta Python tana sarrafa bayanan haske (tsawon hanyoyi, ƙarfi, wuraren hulɗa) don kwaikwayin tsarin soke daidaitawa na na'urar da samar da taswirorin zurfi na ƙarshe da bayanan danye.

3.3 Tasirin Gani da Ake Taimakawa

Kutsawa ta Hanyoyi Da Yawa (MPI): Yana kwaikwayin hasken da ke jurewa nunawa da yawa tsakanin abubuwa kafin ya isa na'urar, babban tushen kuskure a cikin ainihin tsarin ToF.
Abubuwa Masu Watsa Haske/Abubuwa Masu Girman Gaske: Yana lissafin watsawa ƙarƙashin ƙasa da jigilar haske a cikin kayan.
Kurakuran Ruwan Tabarau: Yana ƙirar karkacewa (distortion), vignetting, da sauran lahani na ruwan tabarau waɗanda ke shafar kusurwar shiga da ƙarfin haske akan kowane pixel.
Tushen Haske Mai Faɗi & Da Yawa: Yana ba da damar saitin haske na gaskiya fiye da tushen maki guda.

4. Cikakkun Bayanai na Fasaha da Tushen Lissafi

Kwaikwayon yana ƙirar tsarin haɗin kai (correlation process) a tsakiyar C-ToF. Don mitar daidaitawa $f_{mod}$, siginar da aka karɓa a pixel $(i,j)$ ana haɗa ta da siginonin tunani. Ana cire lokacin (phase) $\phi_{i,j}$ daga samfuran haɗin kai, sau da yawa ta amfani da hanyar samfurin lokaci huɗu (four-phase sampling): $\phi_{i,j} = \arctan\left(\frac{Q_3 - Q_1}{Q_0 - Q_2}\right)$ inda $Q_0$ zuwa $Q_3$ su ne ƙimar haɗin kai a matakan karkatar lokaci na 0°, 90°, 180°, da 270°. OPL da aka kwaikwaya yana shafar waɗannan ƙimar haɗin kai kai tsaye.

5. Sakamakon Gwaji da Nunawa

Takardar ta nuna tsarin akan sauƙaƙan yanayin gwaji na 3D. Manyan sakamako sun haɗa da:

Kwatanta da Gaskiya ta Ƙasa: Taswirar zurfin da aka kwaikwaya ta nuna babban yarda da ƙimar da ake tsammani ta lissafi don hanyoyi kai tsaye.
Samarwar Kayan Kuskuren MPI: Kwaikwayon ya yi nasarar samar da alamu na kuskuren zurfi na halayen kutsawa ta hanyoyi da yawa (MPI), waɗanda galibi ana iya ganin su a matsayin "fatalwa" ko saman da suka karkace a kusurwoyi.
Hoton Tasirin Ruwan Tabarau: Hotunan da aka kwaikwaya sun nuna karkacewa mai radial da vignetting, suna ba da damar binciken tasirin su akan daidaiton zurfi a fagen gani.

Wannan tabbatarwa ya tabbatar da amfanin hanyar don gano da fahimtar abubuwan da ba su dace ba kafin ƙirar samfurin zahiri.

6. Tsarin Bincike: Fahimta ta Asali & Zargi

Fahimta ta Asali

Wannan aikin ba kawai wani kayan aikin kwaikwayi ba ne; gada ce ta dabara tsakanin ƙirar gani mai kyau da gaskiyar rikice-rikice na fahimtar ToF. Ta hanyar goyon bayan tsawon hanyar gani a matsayin madaidaicin ma'auni na kwaikwayi, marubutan sun gano daidai cewa yawancin kurakuran ToF ba hayaniyar lantarki ba ne amma kayan fasaha na tsarin gani na yau da kullun—MPI, watsawa ƙarƙashin ƙasa, kurakuran ruwan tabarau—waɗanda aka shigar da su cikin sigina kafin ta bugi na'urar gano shi. Wannan yana canza mayar da hankali na ingantawa daga ƙirar da'ira ta tsantsa zuwa haɗin ƙirar gani da lantarki gaba ɗaya.

Kwararar Ma'ana

Ma'ana tana da ƙarfi: 1) Yardar cewa jigilar haske na ainihin duniya yana da sarkakiya (bounce da yawa, mai girma). 2) Gane cewa daidaitaccen binciken haske (raytracing) don ƙarfi (kamar a cikin zane-zane na kwamfuta) bai isa ba don fahimtar tushen lokaci (phase). 3) Saboda haka, bincika kuma ka tara tsawon hanyoyin gani, ba kawai ƙarfi ba, ga kowace hanyar haske. 4) Yi amfani da wannan bayanin OPL na zahiri don tafiyar da samfurin haɗin kai/soke daidaitawa. Wannan bututun yana kama da ainihin ilimin kimiyyar lissafi fiye da hanyoyin da suke ƙara tasirin gani azaman tacewa bayan sarrafawa zuwa taswirar zurfi mai kyau.

Ƙarfi & Kurakurai

Ƙarfi: Babban ƙarfin hanyar shi ne gabaɗaya. Ta hanyar raba kwaikwayin gani (Zemax) daga samfurin na'urar (Python), zai iya daidaitawa da nau'ikan ToF daban-daban (D-ToF, C-ToF) har ma da sabbin dabaru kamar ɗaukar hoto na ɗan lokaci (transient imaging), kamar yadda marubutan suka lura. Wannan ya fi sassauƙa fiye da na'urorin kwaikwayi na keɓaɓɓu, na takamaiman na'ura. Taimakon don siffa mai sarkakiya da kayan yana da mahimmanci ga aikace-aikacen mota da na mutum-mutumi inda na'urori ke fuskantar yanayi masu ƙalubale.

Kuskure Mai Muhimmanci: Giwa a cikin ɗaki shine farashin lissafi. Takardar ta ɗan ambaci "sauƙaƙan yanayin gwaji na 3D." Babban amincin binciken haske (raytracing) don miliyoyin haske a cikin yanayi mai yawa, mai yawan bounce yana da tsada sosai don sake zagayowar ƙira. Duk da yake kayan aiki kamar OptiX na NVIDIA sun kawo juyin juya hali ga aikin binciken haske (raytracing), ba a tattauna haɗin kai a nan ba. Bugu da ƙari, samfurin ya bayyana yana aiki galibi a cikin ilimin kimiyyar lissafi na geometric. Don ƙananan na'urorin ToF (misali, a cikin wayoyin hannu), tasirin rarrabuwa (diffraction) da ilimin kimiyyar lissafi na igiyar ruwa (wave optics) a gefunan buɗaɗɗen ruwan tabarau na iya zama mahimmanci, iyakancewa mai kama da waɗanda ake fuskanta wajen ƙirar ƙananan na'urori masu ɗaukar hoto.

Fahimta Mai Aiki

1. Ga Masu Ƙirar Tsarin ToF: Yi amfani da wannan hanyar a cikin lokacin ƙirar gine-gine na farko. Kafin kulle cikakkun bayanai na ruwan tabarau ko tsarin haske, yi kwaikwayi don ƙididdige kasafin kuɗin kuskuren MPI don yanayin da kuke niyya (misali, cikin mota). Wannan zai iya haifar da buƙatu don dabarun mitoci da yawa ko haɓaka algorithms don yaƙar MPI.
2. Ga Masu Haɓaka Algorithms: Wannan na'urar kwaikwayi cikakken dandali ne don samar da manyan, cikakkun bayanai na roba na zahiri don horar da samfuran koyo mai zurfi don cire MPI da sauran kayan fasaha, kama da yadda ake amfani da hanyoyin sadarwa irin na CycleGAN don fassarar hoto zuwa hoto a cikin hangen nesa na kwamfuta. Rashin irin wannan bayanan gaskiya na ainihi, mai yawa, mai lakabin gaskiya shine babban matsalar toshewa.
3. Mahimman Aiki na Gaba: Al'umma dole ne su yi aiki zuwa ga daidaitaccen tsarin kwaikwayi na ToF na buɗe tushe wanda ke daidaita daidaiton zahiri da sauri—watakila ta amfani da filayen haske na jijiyoyi (NeRFs) ko wasu dabarun nunawa masu bambanci don ƙirar samfurin gaba mai sauri, mai koya na samuwar hoton ToF.

7. Hangar Aikace-aikace da Hanyoyin Gaba

Tsarin kwaikwayi yana buɗe hanyoyi don aikace-aikace masu ci gaba da yawa:

Tsarin Kai-da-Kai: Tabbatar da aikin na'urar ToF a farko a cikin matsanancin yanayi (hazo, ruwan sama mai yawa, saman haske) don LiDAR na mota da kewayawar mutum-mutumi.
Binciken Halittu da Lafiya: Ƙirar hulɗar haske tare da nama na ɗan adam don sa ido kan ilimin halittu (misali, bugun zuciya ta hanyar girgiza ƙananan) ta amfani da ka'idojin ToF.
Ƙara/Gaskiyar Gaskiya (AR/VR): Ƙirar ƙananan na'urorin ToF don daidaitaccen bin sawun hannu da taswirar muhalli a cikin kayan kai, kwaikwayin aiki a ƙarƙashin yanayi daban-daban na haske da yanayin kayan.
Auna Masana'antu: Babban kwaikwayi mai daidaito don mutum-mutumin dubawa da ke aiki a cikin yanayi mai nunawa ko cike da tarkace.

Bincike na Gaba yakamata ya mai da hankali kan haɗa ilimin kimiyyar lissafi na igiyar ruwa (wave optics), haɓaka lissafi ta hanyar binciken haske na GPU/na gajimare, da ƙirar hanyar haɗi kai tsaye zuwa samfuran hayaniyar lantarki (misali, hayaniyar harbi, hayaniyar zafi) don ainihin hasashen rabo na sigina zuwa hayaniya (SNR) daga ƙarshe zuwa ƙarshe.

8. Nassoshi

Baumgart, M., Druml, N., & Consani, C. (2018). Procedure Enabling Simulation and In-Depth Analysis of Optical Effects in Camera-Based Time-of-Flight Sensors. The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, XLII-2, 83-90.
Druml, N. et al. (2015). REAL3™ 3D Image Sensor. Infineon Technologies.
Jarabo, A., et al. (2017). A Framework for Transient Rendering. ACM Computing Surveys.
Lange, R. (2000). 3D Time-of-Flight Distance Measurement with Custom Solid-State Image Sensors in CMOS/CCD-Technology. PhD Thesis, University of Siegen.
Remondino, F., & Stoppa, D. (Eds.). (2013). TOF Range-Imaging Cameras. Springer.
Schwarte, R., et al. (1997). A New Electrooptical Mixing and Correlating Sensor: Facilities and Applications of the Photonic Mixer Device (PMD). Proc. SPIE.
Zhu, J.-Y., et al. (2017). Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks. IEEE International Conference on Computer Vision (ICCV). (Nassoshi na CycleGAN don samar da bayanan roba).
NVIDIA OptiX Ray Tracing Engine. (n.d.). An samo daga developer.nvidia.com/optix.