
 
 
 
 
 
 
 
 Umbhali: (1) Yan Long, Electrical Engineering and Computer Science, University of Michigan, Ann Arbor, USA (yanlong@umich.edu); (2) Chen Yan, College of Electrical Engineering, Zhejiang University, Hangzhou, China (yanchen@zju.edu.cn); (3) Shilin Xiao, College of Electrical Engineering, Zhejiang University, Hangzhou, China (bixilin@zju.edu.cn); (4) Shivan Prasad, Electrical Engineering and Computer Science, University of Michigan, Ann Arbor, USA (shprasad@umich.edu); (5) Wenyuan Xu, College of Electrical Engineering, Zhejiang University, Hangzhou, China (wyxu@zju.edu.cn); (6) Kevin Fu, Electrical Engineering and Computer Science, University of Michigan, Ann Arbor, USA (kevinfu@umich.edu). Authors: (1) Yan Long, Electrical Engineering and Computer Science, University of Michigan, Ann Arbor, USA (yanlong@umich.edu); (2) Chen Yan, College of Electrical Engineering, Zhejiang University, Hangzhou, China (yanchen@zju.edu.cn); (3) Shilin Xiao, College of Electrical Engineering, Zhejiang University, Hangzhou, China (bixilin@zju.edu.cn); (4) Shivan Prasad, Electrical Engineering and Computer Science, University of Michigan, Ann Arbor, USA (shprasad@umich.edu); (5) Wenyuan Xu, College of Electrical Engineering, Zhejiang University, Hangzhou, China (wyxu@zju.edu.cn); (6) Kevin Fu, Electrical Engineering and Computer Science, University of Michigan, Ann Arbor, USA (kevinfu@umich.edu). Umbala we-Left I-Abstract kunye ne-Introduction II. Model Threat & background III. I-Webcam I-Peeking Through Glasses IV. Ukukhetha Ukukhetha & Factors V. I-Cyberspace Textual Target Susceptibility VI. Ukukhishwa kwewebhu VII. Ukudibanisa I-III. Izixhobo ezinxulumene IX. Ukuhlaziywa, Ukuvumelana, kunye neengxaki I-Appendix A: Iinkcukacha ze-Equipment I-APPENDIX B: Ukubonisa imodeli ye-angle I-APPENDIX C: Ukusebenza kwe-Video Conference Platform I-Appendix D: Ukuhlolwa kwe-Distortion I-Appendix E: I-Web Textual Goals —Personal video conferencing has become a new norm after COVID-19 caused a seismic shift from in-person meetings and phone calls to video conferencing for daily communications and sensitive business. Video leaks participants’ on-screen information because eyeglasses and other reflective objects unwittingly expose partial screen contents. Using mathematical modeling and human subjects experiments, this research explores the extent to which emerging webcams might leak recognizable textual and graphical information gleaming from eyeglass reflections captured by webcams. The primary goal of our work is to measure, compute, and predict the factors, limits, and thresholds of recognizability as webcam technology evolves in the future. Our work explores and characterizes the viable threat models based on optical attacks using multi-frame super resolution techniques on sequences of video frames. Our models and experimental results in a controlled lab setting show it is possible to reconstruct and recognize with over 75% accuracy on-screen texts that have heights as small as 10 mm with a 720p webcam. We further apply this threat model to web textual contents with varying attacker capabilities to find thresholds at which text becomes recognizable. Our user study with 20 participants suggests present-day 720p webcams are sufficient for adversaries to reconstruct textual content on big-font websites. Our models further show that the evolution towards 4K cameras will tip the threshold of text leakage to reconstruction of most header texts on popular websites. Besides textual targets, a case study on recognizing a closed-world dataset of Alexa top 100 websites with 720p webcams shows a maximum recognition accuracy of 94% with 10 participants even without using machine-learning models. Our research proposes near-term mitigations including a software prototype that users can use to blur the eyeglass areas of their video streams. For possible long-term defenses, we advocate an individual reflection testing procedure to assess threats under various settings, and justify the importance of following the principle of least privilege for privacy-sensitive scenarios. Abstract Ukucinga I. Ukuvula Iividiyo ze-intanethi ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye ziye z Nangona i-videoconferencing ibonelela abantu kunye ne-immersion ye-interactions ye-visual, ibonelela ngempumelelo iinkcukacha ze-text sensitive enokuthintela ukuba ingasetyenziswa ngempumelelo esebenza njenge-participant. i-screen ye-participant ingabandakanyeka iinkcukacha ze-private. I-webcam ye-participant inokufumana le nkcukacha xa ifakwe yi-brass ye-participant kwaye ibonise iinkcukacha kwi-adversary (i-Figure 1). Sinikezela le nentsholongwane njenge-cam peeking attack. Ukongezelela, i-capacity ye-adversary iyaqhubeka ukwandisa kuphela ngexabiso kwi-resolution, i-frame rate, kunye nokunye. Ngoko ke kubalulekile ukufumana iziphumo kunye neengxaki ze-cam peeking kwiimeko ezidlulileyo kunye namaxesha elidlulileyo. Izifundo ezidlulileyo zibonisa ukuba iingxaki efana ne-optical reflection ukusetyenziswa kwezinto ezidlulileyo kwizakhiwo eziholide zibonakalayo, njenge-observing i-teapots kwi-desktop kunye ne-high-end digital single-lens reflex (DSLR) camera kunye ne-telescopes kwi-distance [25], [26]. I-challenge kunye ne-characterization ye-peeking usebenzisa i-webcams ezininzi ezininzi ezininzi, kunjalo, ziquka iimifanekiso ezincinane ze-webcams ezidlulileyo. Iimifanekiso ze-webcams ezincinane zihlanganisa iintlobo ezizodwa zeengxaki, ngoko ke i-shot kunye ne-ISO-noise ngenxa ye- Kukho iintlobo ezininzi zeemithi ezinikezela kwi-reflections ye-optical, kuquka i-text kunye ne-graphics. Thina siphinde i-textual leakage kule nqakraza njengoko kwindawo yokuqala yokuqala yokusebenza kwe-measureable recognizability kunye ne-modelling ye-baseline ye-information leakage, kodwa kunikeza iinkcukacha kwi-leakage ye-informations non-textual ezifana ne-deduction ye-websites ezibonakalisiweyo nge-recognition ye-graphic content kwi-screen. Thina sincoma ukufumana imibuzo emininzi ezilandelayo: Q1: Yintoni iingxaki eziqhelekileyo ezinxulumene ne-camera peeking ability ye-adversary? Q2: Yintoni Ukuphendula i-Q1, sinikezela imodeli ye-mathematics efanelekileyo kodwa efanelekileyo ye-pixel ye-reflection. Le imodeli ibandakanya iimpawu efana ne-camera-screen-resolution kunye ne-glass-screen-distance kwaye ivumela ukuhlaziywa kwinqanaba ye-camera-peeking njengoko ubuchwepheshe ye-camera kunye ne-video zihlabathi. Ngokusebenzisa i-complex-wavelet structural similarity index njenge-metric efanelekileyo ye-reflection-recognizability, sinikezela ukuhlaziywa kwe-semi-quantitative yeempawu ezininzi zonyango, kuquka i-intensity ye-environmental light eyenza i-signal-to-no Ukusabela kweQ2, sinyaniseka iingxaki kwiifoto ze-camera kwaye sincoma ukuguqulwa kwe-super-resolution ye-multi-frame ukuze kwenziwe i-imaging efanelekileyo ukuze kwandise iingxaki. Kule nqakraza idatha ye-brill reflection kwiimeko ze-lab ezisetyenzisiweyo kunye nokuqinisekisa iingxaki ze-reflection ngokusebenzisa abasebenzi be-crowdsourcing kwi-Amazon Mechanical Turk kunye ne-optical character recognition models. I-evaluation ibonisa ngaphezu kwe-75% ngokufanelekileyo kwi-recognition yeengxaki ze-10mm kunye ne-camera 720p Ukuphendula i-Q3, sinxibelelanisa kwi-web text targets ukuvelisa i-benchmark ebonakalayo iingxaki ze-camera ye-peeking. Okokuqala sinxibelelanisa iingxaki ezivela kwimodeli kunye ne-evaluations kwi-web text content ngokufunda iingxaki ezidlulileyo kwi-web text size kunye nokufaka iingxaki kwi-117 i-big-font websites. Emva koko, sinxibelelanisa iingxaki ze-user study kunye ne-20 abalandeli kunye ne-challenge-response game apho umbhali omnye isebenza njenge-adversary yokufunda iinkcukacha ze-HTML eyenziwe ngama-authors. Imiphumo ye-user study ibonisa ukuba iingxaki ze-webcams ze Okugqibela, sinxibelelana iingxaki ezininzi eziluncedo, kubandakanya ukuguqulwa kwe-environmental lighting kunye nokugqiba kwindawo ye-glass kwi-software. Kwakhona sincoma izixazululo ezininzi eziluncedo ngokutsho inkqubo ye-reflection ye-individual kunye nenkqubo ye-minimum privilege. Ngokutsho, inkxaso, inkxaso yokusebenza leyo kuyinto ukunika isiseko se-theoretical kunye ne-benchmark ye-studium yeengxaki ze-camera peeking nge-camera-technology ezintsha kunye ne-development ye-infrastructure ye-video-conferencing. Thina sincoma iingxaki zethu eziphambili: ∙ Umyalelo lethu iingcebiso kunye neengcebiso ezininzi ezidlulileyo ezibonisa iingcebiso zentengiso ze-camera peeking ngokusebenzisa iimodeli ezihlabathi kunye neengcebiso. Le nkcebiso inikeza ukuhlaziywa kwiziphumo ezidlulileyo ezaziwayo ezihlangene neengcebiso zentengiso ze-camera ezihlabathi ezidlulileyo. ∙ Umgangatho we-benchmarking esekelwe kwi-web text targets eyenza ukuguqulwa kweengxaki ze-webcam peeking. I-benchmarking yethu isekelwe kwi-web text design conventions kunye ne-20-participant user study kwi-cams ezidlulileyo ukuze le benchmark ingasetshenziswe kumakhamera emangalisayo kunye neengxaki ezidlulileyo kwiminyaka elandelayo. ∙ I-Analysis yeengxaki zangaphakathi, kubandakanya ukusetyenziswa kwe-software-based blurring filters kunye nokuguqulwa kwe-physical settings kunye neengxaki zangaphakathi zokusetyenziswa ngempumelelo kunye nesiseko se-minimum privilege. I-Analysis yethu ibandakanya i-effectiveness ezininzi kunye ne-methods ye-implementation yeengxaki ezahlukahlukeneyo. 
 
 Oku kunokwenzeka kwi-archiv phantsi kolawulo lwe-ATTRIBUTION-NONCOMMERCIAL-NODERIVS 4.0 INTERNATIONAL. Le nqaku Ngokutsho i-ATTRIBUTION-NONCOMMERCIAL-NODERIVS 4.0 INTERNATIONAL. Zifumaneka kwi-Archiv Zifumaneka kwi-Archiv

Part of HackerNoon's growing list of open-source research papers, promoting free access to academic material.

Read My Stories

undefined's blog

Le audio iveliswe ngolwimi lokuqala lwebali!

Inde kakhulu; Ukufunda

Boost your HackerNoon story @ $159.99! 🚀

Iimpawu zayo ziquka iingcebiso ze-sensitive kwi-video calls

About Author

IZIMVO

ZIJONGE IIMPAWU

ELI NQAKU LINIKEZELWE KU

Related Stories

Light-Mode

Classic

Newspaper

Dark-Mode

Neon Noir

Minty

HN StartUps