Unphotogenic Light Evaluation and Detail of the High-Speed Projection Method to Prevent Secret Photography with Small Cameras Ippei Suzuki, Shinnosuke Ando, Yoichi Ochiai University of Tsukuba, Strategic Research Platform towards Digital Nature Powered by Pixie Dust Technologies, Inc. Session 70 - Projection: Screen Technology

2

3 Introduction Content Protection on the Screen Protecting images projected on a screen is a difficult problem

4 Introduction Content Protection on the Screen Copy Contents is protected by data copy guard techniques People can take contents shown on the screen but Protecting images projected on a screen is a difficult problem

Examples of the content captured by taking the screen 5 *This image is only for illustration purposes, NOT actual result from the video-sharing services.

Visible to Human Eyes Recordable by Cameras ≒ 6

7

8 Introduction Practical Stories of the Proposed Method When you would like to protect this image from secret photography... Situation

9 Introduction Practical Stories of the Proposed Method Solution A: Visible Watermarks / Protection Marks CONFIDENTIAL COPYRIGHT NO PHOTOS

CONFIDENTIAL COPYRIGHT NO PHOTOS 10 Introduction Practical Stories of the Proposed Method Bad Experience for Visitors… Solution A: Visible Watermarks / Protection Marks

Security Check Area 11 Introduction Practical Stories of the Proposed Method Security Visitor Visitor’s Bag Solution B: Confiscate Camera Devices Confiscate Camera Devices

12 Introduction Practical Stories of the Proposed Method Visitor Security Don’t use cameras!!! Solution B: Confiscate Camera Devices Security can orally warn overlooked cameras

13 Introduction Practical Stories of the Proposed Method Visitor’s Pocket Visitor Secret Camera e.g., Pen-Shaped Small Camera They can take photos Solution B: Confiscate Camera Devices

14 Introduction Target in This Study Our Target Kinds of the situation protecting the content Confiscate Camera Devices but contents are secretly captured CONFIDENTIAL COPYRIGHT NO PHOTOS Visible Watermarks not only camera, but also people

15 Introduction Target in This Study Kinds of the camera which criminals may use Our Target “Spy Car Remote Control Camera” by DigitPedia Website (CC BY 2.0, Color Modified) “Sony Xperia Z3 smartphone” by John Jones (CC BY 2.0, Color Modified) “Google Glass” by Kārlis Dambrāns (CC BY 2.0, Background deleted) “Sony Alpha A77” by SkywalkerPL (CC BY 3.0) Not considered in ours Camera Camera Spy Camera Smart Glass Smartphone DSLR

16 Introduction Goal Human eyes can see Cameras can NOT see = Unphotogenic

“Unphotogenic” — Looking non-attractive in photographs or on film — Not appear in photos as they really are in real life 17

18

< Temporal Resolution 19

20 Related Work VRCodes [2] Detection of Imperceptible On-Screen Markers with Unsynchronized Cameras [1] [1]
 [2] Invisible Markers on Screen Sampaio Luiz, Yamada Yoshio, Yamamoto Goshiro, Taketomi Takafumi, Sandor Christian, and Kato Hirokazu. 
 Detection of imperceptible on-screen markers with unsynchronized cameras. IPSJ SIG Notes. CVIM, 2015(64):1–4, jan 2015. Grace Woo, Andy Lippman, and Ramesh Raskar. Vrcodes: Unobtrusive and active visual codes for interaction by exploiting rolling shutter. In Proceedings of the 2012 IEEE International Symposium on Mixed and Augmented Reality (ISMAR), ISMAR ’12, pages 59–64, Washington, DC, USA, 2012. IEEE Computer Society.

21 Afterimage An impression of a vivid sensation retained after the stimulus has ceased

22 Method Approach

23 Method Approach Divide Image Switch in High Refresh Rate

24 Method Theory and Effect of the Proposed System

25 Implementation Projector DLP® LightCrafter™ 4500 EVM Texas Instruments DMD 912 × 1140 diamond pixel 0.45-inch WXGA DMD Resolution Up to 1280 × 800 Light source RGB LED light engine
 with 150 lumen light output Max video speed Up to 120 Hz Max pattern speed Up to 4225 Hz binary pattern
 Up to 120 Hz 8-bit grayscale pattern

26 210 mm 297 mm Projector Camera Screen

27 Evaluation Patterns of division Plasma effect Patterns of division in evaluation Reference: Lode's Computer Graphics Tutorial
 http://lodev.org/cgtutor/plasma.html

28 Result Evaluation from Camera Vision DSLR Smartphone Variable Aperture Fixed Aperture SS 1/15, F1.8 Canon EOS Kiss X4 w/ SIGMA 17-50mm F2.8 EX DC OS HSM All photos were taken by ISO 100, 50mm. Apple iPhone 7 w/ App “ProShot” by Rise Up Games LLC.) All photos were taken by F/1.8, ISO 20, 28mm. SS 1/30, F1.8 SS 1/120, F1.8 SS 1/160, F1.8 **Focal lengths are the corresponding value for a 35-mm Film. SS 1/15, F18 SS 1/30, F13 SS 1/125, F5.6 SS 1/160, F5.6 *Pictographs of shutter speeds and apertures indicate the rough value. Smartphones cannot change the aperture Aperture (F-stop) Shutter speed

29 Method Theory and Effect of the Proposed System Don’t use cameras!!! Combination of the human power and our system Picture which our system applied Visitors cannot take photos of the picture!

Global Shutter Rolling Shutter (popular) 30 Result Evaluation from Camera Vision (other smartphones & cameras) Apple Inc. iPhone 7 G8142; Sony Mobile Communications Inc. Xperia XZ Premium 5040o; TCL Communication Ltd. Alcatel OneTouch Flint PRA-LX2; HUAWEI Technologies Co. HUAWEY nova lite DSC-W110, Sony Corp. Cyber-shot Not affected by the type of the image sensor *Photos are taken with the default application, automatic mode.

31 Application Other Examples than Displaying Divided Images EXIT $10 Theme Park Photo by Ted Murphy (CC BY 2.0) Photo by Ted Murphy (CC BY 2.0) Result Overlay onto the photos sold near the exit of the attraction

32 Application Other Examples than Displaying Divided Images Museum Result Overlay onto the artworks Projector Artwork

33 Discussion & Future Work Robustness Reconstructable from multiple photographs? All parts of the original image are needed

34 Discussion & Future Work Flicker *This video is for illustration purposes. When a saccade occurs on human's eyes or when human blink, people sometimes perceive flicker. Tradeoffs between the strength of the system’s resistance to flicker and an acceptable shutter speed (slowness). Not suitable for extended viewing of media? Optimization of patterns of division?

35 Discussion & Future Work Quality of the Display The brightness of each pixel becomes 1/n of the normal projection. But we need enough brightness of the screen to affect our system into photograph. Normal Ours Time Brighter projectors? N times division 1/N brightness Observed

36

37 Contact: [email protected] Unphotogenic Light Evaluation and Detail of the High-Speed Projection Method to Prevent Secret Photography with Small Cameras Ippei Suzuki, Shinnosuke Ando, Yoichi Ochiai University of Tsukuba, Strategic Research Platform towards Digital Nature Powered by Pixie Dust Technologies, Inc. Questions? Ippei Suzuki Shinnosuke Ando Yoichi Ochiai