模切产品项目经营分析报告(项目总结分析).docx
《模切产品项目经营分析报告(项目总结分析).docx》由会员分享,可在线阅读,更多相关《模切产品项目经营分析报告(项目总结分析).docx(26页珍藏版)》请在文库网上搜索。
1、IDOdesigns a new scheme to “duplicate” the original frame, which we call it watch- only video re-encoding. Instead of directly duplicating the original frame, a sequence of d frames (d ? refresh rate video rate ) for each original frame is carefully produced, such that the visual perception of the n
2、ewly designed frames is same as the original video frame while there is a signifi cant information loss in the pirate video. We have implemented KALEIDOand conducted a comprehensive evaluation over a variety of common mobile cameras. Summary of Results: We test KALEIDOover 30 videos of dif- ferent s
3、tyles, use both LCD monitor and projectors for displaying watch-only videos, and use a variety of smart devices to video- tape the video. We evaluate the video quality of watch-only video and pirate video respectively using both subjective video quality measurement (via extensive survey of 50 audien
4、ces) and objec- tive video quality measurement with a number of different metrics (such as PSNR and SSIM). Our experiments confi rm that KALEI- DOpreserves the high-quality screen-eye channel while reducing the secondary screen-camera channel quality signifi cantly. First, the viewing experience of
5、legitimate audience is not affected: over 90% of the surveyed audiences do not see any quality differences between the original video and watch-only video. The average s- core of the survey is over 4 out of 5, indicating the video quality degradation is almost unnoticeable. Second, the scheme is ver
6、y effective for preventing pirate videotaping: among surveyed au- diences, 96% experience a signifi cant quality drop in the pirate video; and the objective video quality measurement of pirate video also confi rms this observation (PSNR dropped over 60%, SSIM dropped over 40%). Notice that, due to v
7、arious techniques used in reducing the quality of the video, the pirate video actually experi- ences a larger quality degradation when played in real-time than in each of the frames in the pirate video. The rest of paper is organized as follows. In Section 2, we briefl y review the preliminary knowl
8、edge about human vision and color- ing, the video encoding, the video display, and the video-taping. In Section 3, we highlight the design space and principles, and the de- signchallengesandopportunities. Wethenintroduceour KALEIDO for generating watch-only video. We report results from our exten- s
9、ive evaluation of KALEIDOin Section 4. We review the related work in Section 6 and conclude the paper in Section 7. 2.BACKGROUND AND PRELIMINARY In this work, we design a special video type, a watch-only video, i.e. the video can be displayed on common devices and be watched by human with the same v
10、isual quality as the original video, but the pirateversion, capturedbypiratesmobilecameras, willsufferase- vere quality degradation. This is challenging as nowadays mobile devices are equipped with sophisticated cameras which are imita- tion of the human eye. Before presenting our design, we briefl
11、y review the properties of the human eye as the information receiver and the constraints that the display and camera technologies place on the transmission of the light signal. 2.1Characterizing Human Vision Human possess a photopic vision system, which is driven by the cone-cells in the retina. Whe
12、n we see the rich light spectra of ob- jects, different light wavelengths stimulate the three kinds of cone- cells of a viewer in different degrees, providing her perception of distinct colors. Color is usually recognized by the viewer with two aspects: (1) luminance, which is the indication of the
13、“brightness” of the light; (2) chromaticity, which is the property that distinguish- es the composition of the light spectra. Color Description: Various models are designed to quantify hu- man color vision. The commonly used 1931 CIE color spaces are the fi rst defi ned quantitative links between th
14、e physical pure colors (i.e., wavelengths) in the electromagnetic visible spectrum and the physiological perceived colors in human color vision. It convert- s the spectral power distribution of light into the three tristimulus values X,Y,Z. Here Y determines the illuminance (brightness), and X and Z
15、 give chromaticity (hue) at that luminance. The chro- maticity values can be presented in a CIE chromatic diagram as il- lustrated in Fig. 1, where coordinates are defi ned by x = X X+Y +Z and y = Y X+Y +Z. The diagram represents all of the colors visible to the average person. In the rest of this p
16、aper, we use (x,y,Y ) values to describe the chromaticity and illuminance of a specifi c color. SpectralColorAdditiveRule: Thecolorsalonganyline-segment between two points can be made by mixing the colors at the end points, which is called the chromatic additive rule. Specifi cally, if we have two c
17、olored light C1and C2with values (x1,y1,Y1) and (x2,y2,Y2), and mix the two colors by shining them simultaneous- ly, we obtain the mixed color (x,y,Y ) denoted by ? (x,y) = Y1 Y1+Y2(x1,y1) + Y2 Y1+Y2(x2,y2) Y = (Y1+ Y2)/2 (1) The rule shows that, the chromaticity for the mixed color lies on the line
18、 segment joining the individual chromaticities, with the node position on the line segment depending on the relative brightness of the two colors being mixed. Clearly, the combination of colors to produce a given perceived color is not unique. For example, the pair C1C2, C3C4, C5C6in Fig. 1 can each
19、 produce the same color C if combined in the right proportions. TemporalColorAdditivePerception: Whenpeoplewatchtem- poral varying colors, they receive both illuminance change and chromaticity change. When two isoluminant colors alternate at fre- quencies of 25Hz or higher, an observer typically per
20、ceives only one fused color, whose chromaticity is determined based on the chromatic additive rule previously discussed. This may also relate to persistence of vision, the theory where an afterimage is thought 373 (White) RGB Color gamut: subset of colors that can be represented by mixing RGB system
21、 CC1 C2 C3 C4 C5 C6 x E y Figure 1: CIE 1931 chromatic di- agram and color mixture. 1/30 s1/30 s Display Frames (120fps) Captured Frames (Rolling Shutter) Recorded Frames ?30fps? Human Vision Tf tl Time trtetd Figure 2: Color perception by human eyes and image capturing by CMOS cameras. 306090 0.2 0
22、 0.4 0.6 0.8 1.0 Temporal Frequency (Hz) 7 10 30 50 65 Normalized Amplitude Modulation Threshold Figure 3: Flicker regression equation for different display fi eld sizes. to persist for approximately 1 16 of a second on the retina, which is also believed to be the explanation for motion perception.
- 配套讲稿:
如PPT文件的首页显示word图标,表示该PPT已包含配套word讲稿。双击word图标可打开word文档。
- 特殊限制:
部分文档作品中含有的国旗、国徽等图片,仅作为作品整体效果示例展示,禁止商用。设计者仅对作品中独创性部分享有著作权。
- 关 键 词:
- 产品 项目 经营 分析 报告 讲演 呈文 总结