2014 DAC: Catch Your Attention
-
Upload
national-taiwan-normal-university -
Category
Presentations & Public Speaking
-
view
58 -
download
1
description
Transcript of 2014 DAC: Catch Your Attention
Catch Your Attention:Quality-Retaining Power Savingon Mobile OLED Displays
Catch Your Attention:Quality-Retaining Power Savingon Mobile OLED Displays
Chun-Han Lin, Chih-Kai Kang, and Pi-Cheng Hsiu
Research Center for IT Innovation, Academia Sinica
This work was presented at IEEE/ACM DAC 2014
Motivation
•OLED is deemed a promising technology to replace LCD for mobile displays– Many nice features– Black image ( 0 power) vs. White image (2xLCD ≃
power)
2
Low-Power Techniques for OLED Displays•Partial display disabling or dimming
– Darken the contents Impact user perception
•Color remapping– Use other colors instead Not applicable to natural
images
•OLED dynamic voltage scaling– Decrease the supply voltage HW support &
rectangular regions
3
4
Inspired by Human Visual Attention
•Different regions in an image – Receive varying degrees of visual attention– Can tolerate different degrees of image distortion
Quality-Retaining Power Saving
• Image pixel scaling– Scale down the pixel values in regions of any shape
•Technical problems– How to segment an image into regions?– How to determine an appropriate scaling ratio for each
region?
•Contributions– Link human visual attention to OLED power saving– Develop an optimal algorithm w/o accurate power
models– Implement two application scenarios on tablets
5
6
Fast and optimal withoutaccurate OLED power models
Distortion (SSIM)
Analysis
Attention (Itti)
Perception (JND)
Conversion
Optimal Algorithm
7
Experiment Setup
•4 images on Samsung Galaxy Tab 7.7– Different characteristics in terms of luminance and saliency
– Performance Metrics- Execution time (second) and power consumption (watt)
– Comparison- A grid-based approach [DAC’12]
Experiment Results – Impact of Tunable Parameters
• Image Distortion– 38-42% when the distortion
threshold is set at 0.94
8
• Number of regions– 38-42% when the number of
regions is set at 5
Experiment Results – GRID vs. CURA
• Execution time (seconds)
•Power consumption (watts)
•Visual quality – See a video demo
9
GRID CURA
Image Converter 27~219 7.6~8.8
Power-Saving Mode 0.97~4.77 0.72~0.811
GRID CURA
Image Converter 237~648 284~572
Power-Saving Mode 362~797 305~595
*PSM uses Lanczos resampling to scale down the resolution for speedup at a cost of less power saving.
Conclusion
• We introduce visual attention into the quality-retaining power-saving design on mobile OLED displays
• We present CURA to realize the notion and have implemented two application scenarios – Samsung Galaxy Tab 7.7 can save 38-42% OLED power
while retaining visual quality
10
Please come to my poster for more details!