Chapter 4 Principle
4.7 Proposed Algorithm
The proposed algorithm presents a backlight driving technique that concurrently saves power and maintains perceived brightness for liquid crystal displays by scaling the intensity, frequency, and duty cycle of its LED backlights. The algorithm applies the Brücke brightness enhancement effect to compensate the perceived brightness loss while the backlight is modulated temporally.
To give an example, the luminance of a blinking backlight with 50% duty cycle decreases half of luminance. For compensating the loss of luminance, the backlight intensity must be increased by LED driver but that can not save power. For saving power and keeping perceived brightness simultaneously, a new backlight driving method is to decrease frequency without increasing intensity or duty cycle. The human eye can feel brighter when backlight blinking frequency under CFFF so that the same perceived brightness only consumes half of power because of 50% duty cycle and the same intensity.
One requirement for such algorithms is to have a model of perceived brightness enhancement. This model can report to the algorithm observer perceived brightness based on the modulated frequency. Then the power savings can be estimated while the relation between luminance and power consumption is characterized. In chapters, some measurements and psychophysical experiments will be done for verifying the proposed algorithm.
4.8 Summary
Compared with the previous backlight scaling algorithms, the proposed backlight driving scheme can save power and keep perceived brightness independent of the image content. The amount of brightness enhancement can be estimated from the experimental results. In the following chapter, the psychophysical experiments will be done.
Chapter 5
Experimental results and Evaluation
We conducted visual experiments to parameterize the Brücke brightness enhancement effect, Ferry-Porter law and the relationship between favorite display luminance and ambient light.
5.1 Apparatus
To demonstrate our concept, we measured and characterized the power and energy consumption of an Apple iPod®, which is capable of playing MP3 music and MPEG4-compressed video clips. Its major components include a hard-disk, a LED-backlit LCD, a lithium-ion battery, a button/wheel interface, and a video processor, as shown in Figure 5-1.
Figure 5-1: iPod’s backlight consists of four white LEDs in series.
Figure 5-2 shows the power profile of playing a video clip. The spikes in the very beginning occurred when a video clip started to play. The hard disk drive consumed a significant amount of power. The following table lists the power consumed in different states of the 1.8” dual-disk 4,200 rpm hard disk drive Toshiba MK6008GAH.
State Power (mW) Start 1800 Reading, Writing, Seeking 1000 - 1100
Idle 400
Standby, Sleep 7 - 12
Figure 5-2: Power profiling of iPod playing a 323-second video clip.
5.2 Instruments
The luminance and chromaticity coordinates of stimuli are the key properties of system for calibration and evaluation. In order to obtain this information, a colorimeter is necessary.
We chose the Konika-Minolta Chroma Meter CS-200, which utilizes three high-sensitivity silicon photo cells, which were filtered to match the Commission International de I’Eclairage (CIE) standard observer response. Chromaticity coordinates (x,y) and luminance (L) as well as color temperature in Kelvin (K) were calculated from the three cells’ measurements. With this compact reliable color analyzer, the luminance and chromaticity coordinates of stimuli were obtained [Minolta].
The frequency and duty cycle of temporal modulated stimuli were controlled by the function generator (Agilent Technologies 33220A), a 20 MHz synthesized function generator
with built-in arbitrary waveform and pulse capabilities. The Agilent 33220A can generate frequency from 10-6 to 2×107 Hz with 10-6 Hz resolution. The standard waveforms (such as sine or square wave) and arbitrary waveforms can be generated or intermittently. Those arbitrary waveforms can be created by the software, Agilent IntuiLink, using graphic user interface on the PC, and then downloaded into the Agilent 33220A through USB cable. With Agilent 33220A, we can generate arbitrary frequency and duty cycle of temporal modulated stimuli for the LED backlight [Agilent].
The DC power supply, Good Will Instrument PPT-3615G, supplies constant direct current to the LEDs. It has three programmable channels that two of them have adjustable voltage from 0 to 36V as well as adjustable current from 0 to 1.5A. The other has adjustable voltage from 0 to 6V and adjustable current from 0 to 3A [GoodWill b]. Due to the restrictions of PPT-3615G, the serial voltage of LEDs in one series can not exceed 36V as well as the parallel current of LEDs which summed up each series can not exceed 3A.
The power consumption can be calculated by the current and voltage that were measured by digital multimeters. For measuring both the current and voltage simultaneously, two multimeters were needed to measure each power line. For this purpose, the multimeter Good Will Instrument GDM-8246 was chosen [GoodWill a].
(a) (b)
(c) (d) Figure 5-3: The instruments: (a) Konika-Minota Chroma Meter CS-200, (b) Agilent
Technology 33220A, (c) Good Will Instrument PPT-3615G, and (d) Good Will Instrument GDM-8246.
5.3 Luminance and Power Modeling
In Figure 5-4, the power supply supplied the constant voltage and adjustable current to the LED backlight. The voltage was read from the power supply and the current was read from the multi-meter simultaneously. In the meantime, the luminance of LCD panel was measured by the CS-200.
Figure 5-4: The experimental environment setup for luminance and power modeling.
For measuring thirteen samples of luminance and power data, the current was adjusted from 2 to 26 mA with 2 mA increment. For each sample, the voltage and the luminance were recorded at the same time. The power consumption was calculated by multiplying current and voltage and the results are shown in Figure 5-5.
0.00
0.00 10.00 20.00 30.00 40.00 50.00 60.00 70.00
Luminance (cd/m2)
Power (mW)
Figure 5-5: Power vs. luminance of the iPod backlight.
Based on the measurement data, the luminance vs. power relationship can be modeled by the following quadratic function:
10.5760
Where b is backlight luminance.
5.4 Perceptual Flicker Modeling
The conventional psychophysical method of adjustment was used to find the CFFF of three observers in a dark room [Sekuler 1994]. We designed a pair of LED lights for this experiment. The 4cm*5.5cm light was placed 70 cm away from the observer. The light was driven by 50% square waveforms at frequency 20, 25, 30, 35, and 40 Hz. The experimental conditions are illustrated in Figure 5-6.
Figure 5-6: The experimental environment for perceptual flicker modeling.
Each observer was asked to find the CFFF by adjusting the luminance. The results are shown in Figure 5-7. Figure 5-8 was obtained by redrawing Figure 5-7 after replacing the y-axis/x-axis with logarithmic luminance.
0 5 10 15 20 25 30
15 20 25 30 35 40 45
Frequency(Hz) Luminance(cd/m2 )
Geoff Corey Andy
Figure 5-7: Experimental data of the perceptional flicker modeling from three observers.
15
Figure 5-8: Experimental data of the perceptional flicker modeling from three observers as fCFFF vs. log(luminance).
By linearly fitting the data to equation (21), the perceived flicker can be modeled as
.
5.5 Brightness Enhancement Modeling
The brightness enhancement effect was demonstrated by having an observer compare the brightness of two flashes: a test flash and a control flash. The test flash was of variable duration and constant intensity. The control flash was of constant duration and variable intensity. The observer’s task was to adjust the intensity of the control flash such that it matched the brightness of the test flash. Over the course of an experiment, the test flash was presented in a large number of durations, and for each of the durations, the observer adjusted the intensity of the control flash so that it matched the brightness of the test flash. The experimental setup is described in Figure 5-9.
Figure 5-9: Experimental setup for the brightness enhancement effect: An observer saw two circle stimuli simultaneous and adjusted the flashing duration of test light to match the brightness of control light for each variable luminance.
In the experiment of Brücke effect, two 1-degree white LED lights were placed side by side for the observer to match. One was driven by adjustable constant intensity, while the other was driven by 50% square waveforms over a range of 30, 40, and 50 cd/m2. Each observer was asked to match the brightness by adjusting the steady light. The results are shown in Figure 5-10.
0
Figure 5-10: Experimental data of the brightness enhancement effect: brightness vs.
frequency from 3 observers at luminance 30, 40, and 50 cd/m2.
In Figure 5-10, flashes with higher intensity have higher brightness. According to equation (9), the CFFFs are 43, 42, and 41 Hz for 50, 40, and 30 cd/m2, respectively. Beyond 43 Hz, the brightness was about half of the intensity because of the 50% duty cycle. When the frequency approached to zero (DC), the brightness reached about the full intensity. At these low frequencies, the observers could distinct the on-cycles from the off-cycles and chose the high brightness of on-cycles to match. The brightness reached the maximum of about twice intensity around 10 Hz. The frequency range between CFFF and 10 Hz of Figure 5-10 is redrawn as brightness vs. period in Figure 5-11.
Figure 5-11: In the range between CFFF (≅ 42 Hz) and 10 Hz, the relationship between brightness and 1/f can be linearly approximated.
Summarizing the above observations, we can approximate the Brücke effect by:
( .)
5.6 Surrounding Effects
The surround luminance is one of the most important factors in visual sensation. When the user adapts to a dark surround, he/she may dim the backlight to the lowest level and still has the full range of lightness and chroma. In the mean time, a considerable amount of power savings is achieved without any side effect. In literature of display ergonomics such as TCO’03, the luminance ratio of display to surround is recommended to be set between 10:1 and 100:1 [TCO03]. We conducted visual experiments to find the relationship between
favorite display luminance vs. surround luminance. We visited three users in different offices and measured the surround illuminance. The users were asked to perform different tasks at different levels of surround illuminance with their favorite display luminance. If we assume the reflectance of the surround is similar to middle gray (i.e. 18% reflectance, Munsell N5), then the reflected luminance can be estimated as a linear function of illuminance. The results are shown in Figure 5-12.
The tasks included movie watching, web surfing, and text editing. Generally the favorite display luminance increases linearly as the surround illuminance. The movie watching task had much lower display luminance because the display was driven in the direct draw mode.
For the same user, text editing had lower display luminance than web surfing in order to reduce eye strain. The curves have different trends in the bright portion (>100 lux) and dark portion (<100 lux). The reason may be the users switching between the photopic mode (light adapted, cone-dominating) and scotopic mode (dark adapted, rod-dominating).
Figure 5-12: Favorite display luminance vs. surround illuminance of three users on movie watching, web surfing, and text editing.
5.7 Evaluation
For evaluating the proposed algorithm, the prototype consists the four blocks as shown in Figure 5-13. In the head of the system block, the photo-sensor module detects the ambient light and provided the FPGA module the ambient light levels. Next, the FPGA module determined what kind of environments was located due to the information of ambient light levels. The LED driver drove the LED backlight module by the driving scheme that was provided from the FPGA module. The backlight emitted the light depending on the proposed algorithm implemented in the FPGA module.
0001 0001
Photo-sensor FPGA LED Driver Backlight
Figure 5-13: The system block diagram consisted a photo-sensor module, FPGA module, LED driver, and LED backlight module.
For the brightness of a steady light of luminance L, we blinked the backlight with the following duty cycle, magnitude and frequency instead:
%
Based on equation (23), the magnitude and frequency of the backlight could be reduced while keeping the same brightness. The power consumption was estimated as listed in Table 5-1. In the last two columns, RL represents the brightness ratio of L’50% to L (RL= L’50%/L) and RP represents the power consumption ratio of PL’50% to PL (RP=PL’50%/PL) while L’50% is perceived brightness at f* with 50% duty cycle.
Table 5-1: The brightness enhancement ratio and power consumption ratio were listed by different driving frequency:
L (nit) PL(mW) fCFFF (Hz) f* (Hz) L'50% (nit) PL'50% (mW) RL RP
32.5 133.6 41.6 40.0 33.1 137.1 102% 97%
32.5 133.6 41.6 35.0 35.4 150.3 109% 89%
32.5 133.6 41.6 30.0 38.5 168.6 118% 79%
32.5 133.6 41.6 25.0 42.7 195.4 132% 68%
32.5 133.6 41.6 20.0 49.2 238.1 151% 56%
32.5 133.6 41.6 15.0 59.9 316.1 184% 42%
32.5 133.6 41.6 10.0 81.3 497.9 250% 27%
When the baseline backlight power is 100%, the proposed technique can decrease power consumption to 68% while preserving the same brightness at the cost of flickering. When the baseline backlight brightness was 100%, the proposed technique can increase the brightness up to 132% while keeping the same power consumption at the cost of flickering. The power saving ratio decreases and the brightness enhancement increases with decreased f*. However, the increase of brightness enhancement becomes less due to Broca-Sulzer effect, which has a peak of brightness enhancement at 50ms duration as well as 10 Hz with 50% duty cycle.
The visual effects of different techniques were shown as Figure 5-14. In Figure 5-14 (a), higher gray level pixels are over saturated when Choi’s algorithm was used. In Figure 5-14 (b), the tone of image was changed when Iranli’s algorithm was used. In Figure 5-14 (c) when Cheng’s algorithm was used, both higher and lower gray level pixels was cut but better than (a). In the figure (a), (b), or (c), more power savings lead to more distortion. However, applying the proposed algorithm did not cause spatial distortion but flickering in the temporal domain.
(a) (b)
(c) (d) Figure 5-14: Simulated visual effects of (a) Choi, (b) Iranli, (c) Cheng, and (d) the
proposed algorithm without showing the flickering.
5.8 Summary
The psychophysical experiments had been performed and the relationship between perceived brightness and modulated light frequency had been modeled based on its results. By using this model, the power savings can be estimated. The proposed algorithm can reduce power consumption down to 68% while preserving the same brightness or can increase brightness up to 132% at the cost of flickering.
Chapter 6
Conclusions and Future Direction
6.1 Conclusions
The temporal properties of human vision system such as Broca-Sulzer effect and Brücke-Bartley effect have been reviewed and applied to the low-power LCD. For modeling the relationship between perceived brightness and modulated light frequency, psychophysical experiments have been performed. According to the experimental results, a novel backlight driving technique for liquid crystal displays has been presented. By scaling the intensity, frequency, and duty cycle of the backlight, this technique not only increases the perceived brightness but also prolongs the service time of rechargeable batteries. A great amount of energy can be saved at the cost of flickering. The Brücke brightness enhancement effect from temporal properties of vision system was employed in customizing LED backlight modules.
6.2 Future Direction
Although the preliminary results are encouraging, this study is still in its infancy. Hence, the future direction may extend the temporal domain to the spatial domain through spatial-temporal contrast sensitivity surface and include developing a metric for measuring the flickering, refining the visual experiments, and FPGA implementation. Furthermore, the brightness enhancement of temporal properties of vision is applicable to the emergency light or traffic light.
References
[Agilent] Agilent Technology, 33220A Function Generator.
[Anandan 2006] M. Anandan, ”LED Backlight for LCD/TV Monitor: Issues that Remain”, SID06’ DIGEST, pp. 1509-1512, 2006.
[Anstis 1983] S. M. Anstis and P. Cavanagh, “A minimum motion technique for judging equiluminance,” in Color Vision: Physiology and Psychophysics, J. D. Mollon and L. T.
Sharpe, eds., pp. 155-166, 1983.
[Arend 1994] L. Arend, J. Lubin, J. Gille, J. Larimer, “Color breakup in sequentially scanned LCDs,” SID 94 Digest, pp. 201-204, 1994.
[Arima] Arima Optoelectronics Corporation, 7F, No.349, Sec. 2, Renhe Rd., Dashi, Taoyuan 335, Taiwan (R.O.C.), 5W RGGB LED Emitter.
[Broca 1902] A. Broca, D. Sulzer, “La sensation lumineuse en fonction du temps,” Academie Des Sciences, pp. 831-835, 1902.
[Cheng 2004] W-C. Cheng and M. Pedram, “Power minimization in a backlit TFT-LCD display by concurrent brightness and contrast scaling,” Proc. of Design Automation and Test in Europe, pp. 252-259, 2004.
[Cheng 2006] W-C. Cheng, C-F. Hsu, and C-F. Chao, "Temporal vision-guided energy minimization for portable displays," Proc. of International Symposium on Low Power Electronics and Design, pp.89-94, 2006.
[Choi 2004] N. Chang, I. Choi, H. Shim, “DLS: Dynamic backlight luminance scaling of liquid crystal display,” IEEE Tran. VLSI, Vol. 12, No. 8, pp. 837-846, 2004.
[ConoScope] http://www.autronic-melchers.com/
[Farrell 1987] J. E. Farrell, B. L. Benson, C. R. Haynie, “Predicting flicker thresholds for video display terminals,” Proceedings of the SID, Vol. 28, pp. 449-453, 1987.
[GoodWill a] Good Will Instrument Co., Ltd., No. 7-1, Jhongsing Rd., Tucheng, Taipei, 236, Taiwan (R.O.C.), GDM-8246 multi-meter.
[GoodWill b] Good Will Instrument Co., Ltd., No. 7-1, Jhongsing Rd., Tucheng, Taipei, 236, Taiwan (R.O.C.), PPS-3635G Programmable DC power supply.
[Hecht 1933] S. Hecht, and C. D. Verrijp, “Intermittent stimulation by light. III. The relation between intensity and critical fusion frequency for different retinal locations,” J. Gen. Physiol., Vol. 17, pp. 251-268, 1933.
[Huang 2000] H. C. Huang, P. W. Cheng, and H. S. Kwok, “On the Minimization of Flicker and Image Retention in Silicon Light Valves,” SID 00 Digest, pp. 248-251, 2000.
[Hunt 1995] R. W. G. Hunt, Measuring Colour, 3rd Ed., Fountain Press, England, 1998 [Iranli 2005] A. Iranli, H. Fatemi, and M. Pedram, “HEBS: Histogram equalization for backlight scaling,” Proc. of Design Automation and Test in Europe, pp. 346-351, 2005.
[Kelly 1961] D. H. Kelly, “Visual responses to time-dependent stimuli. I. Amplitude sensitivity measurements,” J. Opt. Soc. Am., Vol. 51, pp. 422-429, 1961.
[Kelly 1966] D. H. Kelly, “Frequency doubling in visual responses,” J. Opt. Soc. Am., Vol. 56, pp. 1628-1633, 1966.
[Kelly 1972] D. H. Kelly, “Adaptation effects on spatio-temporal sine-wave thresholds,”
Vision Res., Vol. 12, pp. 89-101, 1972
[Kelly 1974] D. H. Kelly, “Spatio-temporal frequency characteristics of color-vision mechanisms,” J. Opt. Soc. Am., Vol. 64, pp. 983-990, 1974.
[Kelly 1977] D. H. Kelly, D. van Norren, “Two-band model of heterochromatic flicker,” J.
Opt. Soc. Am., Vol. 67, pp. 1081-1091, 1977.
[Kelly 1979] D. H. Kelly, “Motion and vision. II. Stabilized spatio-temporal threshold surface,” J. Opt. Soc. Am., Vol. 69, No. 10, pp. 1340-1349, 1979.
[Kelly 1981] D. H. Kelly, “Disappearance of stabilized chromatic gratings,” Science, Vol. 214, pp. 1257-1258, 1981.
[Kelly 1983] D. H. Kelly, “Spatiotemporal variation of chromatic and achromatic contrast thresholds,” J. Opt. Soc. Am., Vol. 73, No. 6, pp. 742-750, 1983.
[Kelly 1994] D. H. Kelly, Visual Science and Engineering, Marcel Dekker, 1994.
[Koma 2000] N. Koma, T. Miyashita, T. Uchida, and N. Mitani, “Color field sequential LCD using an OCB-TFT-LCD,” SID 00 Digest, pp. 632-635, 2000.
[Laird 2006] J. Laird, M. Rosen, J. Pelz, E. Montag, and S. Daly, “Spatio-velocity CSF as a function of retinal velocity using unstabilized stimuli,” Proc. of SPIE-IS&T Electronic Imaging, Vol. 6057, 605705, 2006.
[Lee 2005] S. H. Lee, S. W. Jang, E. S. Kim, S. H. Lee, and K. I. Sohng, “The design of an LCD system based on the quantitative analysis for enhancement of the motion picture quality,” IEICE Trans. ELECTRON., Vol. E88-C, No. 11, pp. 2094-2098, 2005.
[Lee 2006] J. W. Lee et al., “Amorphous silicon based 40” LCD TV using ultra fast OCB mode,” SID 06 Digest, pp. 1950-1953, 2006.
[Liu 2005] J. Liu, B. A. Wandell, “Specializations for chromatic and temporal signals in human visual cortex,” Journal of Neuroscience, Vol. 25, pp. 3459-3468, 2005.
[Minolta] Konika-Minolta, Chroma Meter CS-200.
[Moses 1987] R. A. Moses, and W. M. Hart, Jr., Adler’s Physiology of the Eye, 8th Ed., Washington, DC: The C. B. Mosby Company, 1987.
[Post 1997] D. L. Post, P. Monnier, and C. S. Calhoun, “Predicting color breakup on
field-sequential displays,” Proceedings of SPIE: Head-Mounted Displays II, 3058, pp. 57-65, 1997.
[Salters 2001] B. Salters, R. van Dijk, “Reduction of large area flicker in plasma display panels,” SID 01 Digest, pp. 1098-1101, 2001.
[Schwartz 1994] S. H. Schwartz, Visual Perception: A Clinical Orientation, Appleton &
Lange, 1994.
[Sekuler 1994] R. Sekuler and R. Blake, Perception, 3rd Ed., NY: McGraw-Hill, 1994.
[Silverstein 2005] L. D. Silverstein, “STColor: Hybrid spatial-temporal color synthesis for enhanced display image quality,” SID 05 Digest, pp. 1112-1115, 2005.
[SiTI] SiTI, No. 9-7F-1, Prosperity Rd. I, Science-Based Industrial Park, Hsinchu, 300, Taiwan (R.O.C.), DD311 High Constant Current LED Driver.
[Sun 2006] P. L. Sun, “A simple approach to reduce color breakup for field-sequential color LCDs,” Taiwan Display Conference 2006, pp. 314-317, 2006.
[Swanson 1987] W. H. Swanson, T. Ueno, V. C. Smith, and J. Pokorny, “Temporal modulation sensitivity and pulse-detection thresholds for chromatic and luminance perturbations,” J. Opt.
Soc. Am. A, Vol. 4, pp. 1992-2005, 1987.
[TCO03] http://www.tcodevelopment.com/
[Uchida 2006] T. Uchida et al., “Color imaging and display system with field sequential OCB LCD,” SID 06 Digest, pp. 116-169, 2006.
[Varner 1984] D. Varner, “Temporal sensitivities related to color theory,” J. Opt. Soc. Am. A, Vol. 1, pp. 474-481, 1984.
[ViewSonic] ViewSonic, No. 381, Brea Canyon Rd., Walnut, California, 91789, USA, VX912 LCD Monitor.
[Wagner 1972] G. Wagner, and R. M. Boynton, “Comparison of four methods of heterochromatic photometry,” J. Opt. Soc. Am., Vol. 62, pp. 1508-1515, 1972.
[Wandell 1995] B. A. Wandell, Foundations of Vision, Sinauer Associates, 1995.
[Wyszecki 1982] G. Wyszecki, W. S. Stiles, Color Science, 2nd Ed., John Wiley and Sons, 1982.
[Yamagishi 2006] N. Yamagishi, Y. Tanabe, H. Ishibashi, and H. Yamagami, “Moving picture simulations with OCB for image-blurring-free LC TVs,” SID 06 Digest, pp. 1804-1807, 2006.
[YEH 1999] P. Yeh, and C. Gu, Optics of Liquid Crystal Displays, John Wiley & Sons, New York, 1999.
[Zwanenburg 2004] M. J. Zwanenburg et al., "Highefficiency LEDs for LCD backlights,"
SID 04 Digest, pp. 1222-1225, 2004.
Vita
A. GENERAL INFORMATION
A. GENERAL INFORMATION