TV 1,2,a) 1 2 2015 1 26, 2015 5 21 Augmented TV TV AR Augmented Reality 3DCG TV Estimation of TV Screen Position and Rotation Using Mobile Device Hiroyuki Kawakita 1,2,a) Toshio Nakagawa 1 Makoto Sato 2 Received: January 26, 2015, Accepted: May 21, 2015 Abstract: We have studied on a new TV service, named augmented TV, which is able to augment representation of TV programs beyond the TV screen. We have proposed a system in which animated 3DCG content interlocked with TV programs is overlaid on live video from the mobile device camera in the mobile device screen by augmented reality techniques. In the system, the representation of having a TV character coming out of the screen can be provided. To achieve such a representation, it is needed always accurately to estimate position and rotation of the TV in the mobile device. We have proposed an estimation method using the camera and the gyro sensor. This paper shows the details of our method and that it is effective by experiments using the demo content. Keywords: augmented reality, 3DCG, augmented TV, position & rotation estimation, gyro sensor 1. 60 TV 16:9 2 1 1 NHK NHK Science & Technology Research Laboratories, Setagaya, Tokyo 157 8510, Japan 2 Tokyo Institute of Technology, Yokohama, Kanagawa 226 8503, Japan a) kawakita.h-dq@nhk.or.jp Augmented TV [1], [2], [3] Augmented TV Augmented TV Augmented TV TV AR Augmented Reality [4] c 2015 Information Processing Society of Japan 61
1 Augmented TV Fig. 1 Service model of augmented TV. 1 TV TV 3DCG TV TV 1 TV 3DCG 3DCG TV 3DCG TV Augmented TV 1 [5] [1] TV [2], [3] 2. TV 2.1 [1] TV Augmented TV TV 2 TV Fig. 2 Cue to detect TV screen. TV [4] 2 (a) TV TV TV TV TV TV TV [1] TV 2.2 2.1 2K 4K 30 fps 60 fps AR PC 3DCG c 2015 Information Processing Society of Japan 62
TV 2.1 TV TV TV TV Fig. 3 3 Flow chart of the estimation. 3. 3.1 TV TV TV [6] TV Augmented TV TV TV TV TV 2(a) TV 2(b) 2 TV 3.2 TV TV TV 3 TV TV 2 TV TV 3 TV 3 Perspective-n-Point PnP 4 4 OpenCV [7] cv::solvepnp() 3.3 3 1 TV c 2015 Information Processing Society of Japan 63
1 Table 1 Estimation method. 3.4 TV [1] TV ON/OFF 3.5 3 3.6 3.5 TV 3.6.1 1 N O(N) O(N 2 ) 4 Fig. 4 Search path of TV frame. [5], [8], [9], [10] [8] 1/100 [8] [9] [9] 61% 93% 3 [5], [10] TV TV 3.6.2 4 (1) 3.5 (2) c 2015 Information Processing Society of Japan 64
Fig. 6 6 2 Rotation correction at detection of 2 vertices. 5 Fig. 5 Classification of captures of TV frame. (7) (3) (2) (7) (4) (3) (3) 8 2 (5) (4) (3) (4) (6) 3 (3) 2 2 3 3.5.3 (4) (7) (7) (1) (2) 3.6.3 TV 3.6.2 4 TV 2 3 5 3 2 3.6.3.1 3 3 3.6.2 2 5 4 3.6.3.2 2 2 3 2 A) B) TV A 6 A 2 M 2 3 4 2 4 A B A B 2 c 2015 Information Processing Society of Japan 65
TV 3DCG TV TV TV 6 2 A TV 1 A B 2 A B 4. TV 2 4.1 3.6 7 CG Augmented TV CPU 1ms 10 1 OpenCV cv::houghlinesp() 8 OpenCV cv::canny() 6.8 ms 30.2 ms 10 2 30 fps 1 33 ms 1 2ms 4.2 2 2 TV TV Table 2 2 Specification and parameters of implementation. Fig. 7 7 Processing time of the frame recognition. Fig. 8 8 Example of the results of Hough transformation. c 2015 Information Processing Society of Japan 66
5.2 TV 2 4 5. 5.1 3.6.1 O(N) 2K 4K 640 480 px 3 6 2ms 6ms 12 ms 60 fps 16 ms 2.2 Augmented TV HMD Head Mount Display [11] HMD [11] 4.1 2 5.2 4.2 3 [10] TV / TV 0.5 2.0 4.2 TV TV TV TV 3.6.2 6. c 2015 Information Processing Society of Japan 67
9 Fig. 9 NHK State of NHK Science Stadium. 11 3DTV Fig. 11 Model image of search results by a word 3DTV. 10 Fig. 10 Schematic diagram of the demo content. NHK [12] 2 9 10 1 16 2 TV 5 TV TV 150 2 2 TV TV TV TV TV TV 7. Augmented TV TV 50 [13] CM [14] TV TV 3DTV 3DTV 11 3DTV TV 3DTV Augmented TV TV 5.1 TV TV 8. Augmented TV TV TV c 2015 Information Processing Society of Japan 68
TV Augmented TV TV TV 1989 NHK 2012 [1] Augmented TV TV Vol.19, No.3, pp.319 328 (2014). [2] TV 2014 13 1 (2014). [3] TV 2014 4 12 (2014). [4] ARToolKit Vol.101, No.652, pp.79 86 (2002). [5] AR Visual SyncAR 41 R6-2 (2013). [6] Klein, G. and Murray, D.: Parallel Tracking and Mapping for Small AR Workspaces, 6th IEEE and ACM International Symposium on Mixed and Augmented Reality (ISMAR), pp.1 10 (2007). [7] OpenCV ver2.4.10, available from http://opencv.jp. [8] i D-II, Vol.J88, No.6, pp.1035 1046 (2005). [9] FIT2005, J-027 (2005). [10] Vol.106, No.351, pp.1 6 (2006). [11] BP pp.5 33 (2014). [12] NHK http://www.nhkp.co.jp/event/detail.php?id=423. [13] (1965). [14] (1998). 1973 3 1978 3 4 1986 3 VR SPIDAR VR 2004 NHK NHK 2007 c 2015 Information Processing Society of Japan 69