ActionScript3.0 1 1 YouTube Flash ActionScript3.0 Face detection and hiding using ActionScript3.0 for streaming video on the Internet Ryouta Tanaka 1 and Masanao Koeda 1 Recently, video streaming and video sharing using Flash technology such as YouTube are increasingly prevalent in the Internet. However, most of the streaming video are created without any permission and it is deeply concerned about the intrusion of privacy and the leaking of personal information. Such a background, we are targeting the protection of the face privacy in the streaming videos. In this research, face detection and hiding system for flash by using ActionScript3.0 was developed, and we conducted feasibility experiments. 1. YouTube, Google Video, Live Leak, Veoh, 9 1 Osaka Electro-Communication University, Faculty of Information Science and Arts 7 1) 1 2000 Adobe Flash Video(FLV,F4V) Flash Video FlashPlayer6 Flash Video HTML Windows Media Video QuickTime Flash Video web 2)3) 4)5) Adobe Flash Video ActionScript3.0. 2. 2.1 ActionScript ActionScript ECMA-262 Flash Player ActionScript 1.0 2.0 3.0 ActionScript Flash Player Flash Player Flash Player9 1 c 2010 Information Processing Society of Japan
ActionScript1.0 3.0 Flash Player 8 ActionScript3.0 ActionScript Flash Video ActionScript.swf swf FlashPlayer AVM(Actionscript Virtual Machine) Windows Mac AVM Flash Plyaer ActionScript Flash Video AVM AVM1 AVM2 AVM2 ActionScript3.0 ActionScript1.0 2.0 FlashPlayer9 AVM1 AVM2 AVM2 Flash Player PC 2.2 2 web HTTP YouTube 3.1 ( 1 ) NetConnection ( 2 ) NetConnection connect() 6) null 3. 1 1 Fig. 1 Flowchart 2 c 2010 Information Processing Society of Japan
( 3 ) NetStream NetConnection NetStream ( 4 ) Video Video ( 5 ) Video attachnetstream() 6) NetStream ( 6 ) NetStream Play ( 7 ) addchild() 6) Video ( 8 ) RGB Video BitmapData draw() ( 9 ) draw() BitmapData 3.2 RGB HSV 1 HSV G B 60 + 0 if MAX = R MAX MIN B R H = 60 + 120 if MAX MIN MAX = G (1) R G 60 + 240 if MAX = B MAX MIN H+ = 360 if H < 0 (2) MAX MIN S = MAX V = MAX 3.3 HSV H H (10 60) 3.4 0 9 ActionScript3.0 threshold 1 ConvolutionFilter Table 1 matrix used in ConvolutionFilter 1 1 1 1 1 1 1 1 1 6) 1 1 9 9 f (f(i, j)) U V W (f(i, j)) (g(i, j)) 3 U 1 V 1 g(i, j) = f(i + x U/255, j + y V/255) w(x, y) (3) w(x, y) x=0 y=0 1 U = 3, V = 3 3 3 9 255 0 9 3.5 2 ActionScript3.0 floodfill 6) floodfill (x, y) 3.6 2 2 1 200 2 (x + 20, y 10) 60% (x, y) 3 c 2010 Information Processing Society of Japan
3.7 OS CPU Flash Player swf 2 Table 2 Video information fps 320 240[pixel] 49[sec] 30[fps] 3 PC Table 3 Camera, PC s spec Microsoft Windows XP Professional SP3 Intel Core i7 860 @ 2.80[GHz] 3[GB] Var.10,0,42,34 Logicool 2-MP Portable Webcam C905m 138.7 [Kbyte] (x, y) x y 10 (x, y) (x, y) 10*10 4. swf HTML HTML 1 2 3 2 PC 3 4.1 2 4.2 2 3 [sec] 4 [sec] 5 6 200 60% 2 5. ActionScript3.0 3 Flash Video 1) ( 5 )-Yahoo! http://www.yahoo-vi.co.jp/research/100112.html 2) JPEG2000 ROI 4 c 2010 Information Processing Society of Japan
(a)0[sec] (b)5[sec] (c)10[sec] (d)15[sec] (e)20[sec] (f)25[sec] (g)30[sec] (h)35[sec] (i)40[sec] (j)45[sec] 2 Fig. 2 Original image Vol.108, No.86, pp. 27 32 (2008.6) 1235 1244 (1990). 3) Mitsuji MUNEYASU, Shuhei ODANI, Yoshihiro KITAURA and Hitoshi NAMBA: An Implementation of Privacy Protection for a Surveillance Camera Using ROI Coding of JPEG2000 with Face Detection, IEICE Transactions on Fundamentals of Electronics, Communications and Computer Sciences, Vol.E99, No.11, pp.2858 2861 (2009). 4) : Vol.24, No.49, pp.1 6 (2000.9). 5) : Vol.56, No.12, pp.1980 1988 (2002). 6) Adobe Flex 3.2 http://livedocs.adobe.com/flex3 jp/langref/inde x.html. 7) : Vol.71, No.2, pp.449 450 (2009.3). 5 c 2010 Information Processing Society of Japan
情報処理学会研究報告 (a)0[sec] (b)5[sec] (c)10[sec] (d)15[sec] (e)20[sec] (f)25[sec] (g)30[sec] (h)35[sec] (i)40[sec] (j)45[sec] 処理結果 Processing Results 図3 Fig. 3 6 c 2010 Information Processing Society of Japan
label_bf label_af FPS z ikken1 14 bf_label 5 af_label 8 7.9 fps 12 4 7.8 10 7.7 3 7.6 label 8 label FPS 7.5 6 2 7.4 4 2 1 7.3 7.2 7.1 0 0 5 10 15 20 25 30 35 40 45 50 time[sec] 4 : Fig. 4 Experiment:Number of labels before Face detection processing 0 0 5 10 15 20 25 30 35 40 45 50 time[sec] 5 : Fig. 5 Experiment:Number of labels before Face detection processing 7 0 5 10 15 20 25 30 35 40 45 50 time[sec] 6 :,fps Fig. 6 Experiment:Number of labels before processing 7 c 2010 Information Processing Society of Japan