The power consumption of an OLED (Organic Light-Emitting Diode) display depends on the color distribution of the image contents. Previous studies predicted OLED display power using a pixel color-based linear model. In recent mobile devices, however, the relationship between a color and its power consumption has become complicated, due to the various options for display settings in the device and to device diversity. This makes it hard to predict the power consumption of a display accurately with the conventional linear model. In this paper, we propose a technique to automatically generate an OLED display power model optimized for a specific target device, as well as for each display mode, thus generating an accurate power model effectively. The technique automatically learns the power model through the RGB value of each pixel and its power consumption using Support Vector Regression (SVR)-based supervised learning. We evaluated the power models for recent smartphone models, and the results show that, on average, the accuracy of the pixel modeling is about 99% for the device models and display modes used in the experiment. Furthermore, the power consumption for real images is estimated with an accuracy of about 95%, on average.
Bibliographical noteFunding Information:
This research was supported by Samsung Research Funding Center of Samsung Electronics under Project Number SRFC-TB1503-02 and National Research Foundation of Korea (NRF) grant funded by the Korean government, Ministry of Education, Science and Technology under Grant NRF-2017M3C4A7083677 .
© 2018 Elsevier B.V.
All Science Journal Classification (ASJC) codes
- Information Systems
- Hardware and Architecture
- Computer Science Applications
- Computer Networks and Communications