Energy characterization and optimization of image sensing toward continuous mobile vision

Robert LiKamWa, Bodhi Priyantha, Matthai Philipose, Lin Zhong, Paramvir Bahl

Research output: Chapter in Book/Report/Conference proceedingConference contribution

90 Citations (Scopus)

Abstract

A major hurdle to frequently performing mobile computer vision tasks is the high power consumption of image sensing. In this work, we report the first publicly known experimental and analytical characterization of CMOS image sensors. We find that modern image sensors are not energy-proportional: energy per pixel is in fact inversely proportional to frame rate and resolution of image capture, and thus image sensor systems fail to provide an important principle of energy-aware system design: trading quality for energy efficiency. We reveal two energy-proportional mechanisms, supported by current image sensors but unused by mobile systems: (i) using an optimal clock frequency reduces the power up to 50% or 30% for low-quality single frame (photo) and sequential frame (video) capturing, respectively; (ii) by entering low-power standby mode between frames, an image sensor achieves almost constant energy per pixel for video capture at low frame rates, resulting in an additional 40% power reduction. We also propose architectural modifications to the image sensor that would further improve operational efficiency. Finally, we use computer vision benchmarks to show the performance and efficiency tradeoffs that can be achieved with existing image sensors. For image registration, a key primitive for image mosaicking and depth estimation, we can achieve a 96% success rate at 3 FPS and 0.1 MP resolution. At these quality metrics, an optimal clock frequency reduces image sensor power consumption by 36% and aggressive standby mode reduces power consumption by 95%.

Original languageEnglish (US)
Title of host publicationMobiSys 2013 - Proceedings of the 11th Annual International Conference on Mobile Systems, Applications, and Services
Pages69-81
Number of pages13
DOIs
StatePublished - 2013
Externally publishedYes
Event11th Annual International Conference on Mobile Systems, Applications, and Services, MobiSys 2013 - Taipei, Taiwan, Province of China
Duration: Jun 25 2013Jun 28 2013

Other

Other11th Annual International Conference on Mobile Systems, Applications, and Services, MobiSys 2013
CountryTaiwan, Province of China
CityTaipei
Period6/25/136/28/13

Fingerprint

Image sensors
Electric power utilization
Computer vision
Clocks
Pixels
Image registration
Energy efficiency
Systems analysis

Keywords

  • Computer vision
  • Energy efficiency
  • Energy proportionality
  • Image sensor
  • Mobile systems

ASJC Scopus subject areas

  • Computer Networks and Communications
  • Computer Science Applications

Cite this

LiKamWa, R., Priyantha, B., Philipose, M., Zhong, L., & Bahl, P. (2013). Energy characterization and optimization of image sensing toward continuous mobile vision. In MobiSys 2013 - Proceedings of the 11th Annual International Conference on Mobile Systems, Applications, and Services (pp. 69-81) https://doi.org/10.1145/2462456.2464448

Energy characterization and optimization of image sensing toward continuous mobile vision. / LiKamWa, Robert; Priyantha, Bodhi; Philipose, Matthai; Zhong, Lin; Bahl, Paramvir.

MobiSys 2013 - Proceedings of the 11th Annual International Conference on Mobile Systems, Applications, and Services. 2013. p. 69-81.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

LiKamWa, R, Priyantha, B, Philipose, M, Zhong, L & Bahl, P 2013, Energy characterization and optimization of image sensing toward continuous mobile vision. in MobiSys 2013 - Proceedings of the 11th Annual International Conference on Mobile Systems, Applications, and Services. pp. 69-81, 11th Annual International Conference on Mobile Systems, Applications, and Services, MobiSys 2013, Taipei, Taiwan, Province of China, 6/25/13. https://doi.org/10.1145/2462456.2464448
LiKamWa R, Priyantha B, Philipose M, Zhong L, Bahl P. Energy characterization and optimization of image sensing toward continuous mobile vision. In MobiSys 2013 - Proceedings of the 11th Annual International Conference on Mobile Systems, Applications, and Services. 2013. p. 69-81 https://doi.org/10.1145/2462456.2464448
LiKamWa, Robert ; Priyantha, Bodhi ; Philipose, Matthai ; Zhong, Lin ; Bahl, Paramvir. / Energy characterization and optimization of image sensing toward continuous mobile vision. MobiSys 2013 - Proceedings of the 11th Annual International Conference on Mobile Systems, Applications, and Services. 2013. pp. 69-81
@inproceedings{3c0646087d1d4ddbaf336b93a8c557e2,
title = "Energy characterization and optimization of image sensing toward continuous mobile vision",
abstract = "A major hurdle to frequently performing mobile computer vision tasks is the high power consumption of image sensing. In this work, we report the first publicly known experimental and analytical characterization of CMOS image sensors. We find that modern image sensors are not energy-proportional: energy per pixel is in fact inversely proportional to frame rate and resolution of image capture, and thus image sensor systems fail to provide an important principle of energy-aware system design: trading quality for energy efficiency. We reveal two energy-proportional mechanisms, supported by current image sensors but unused by mobile systems: (i) using an optimal clock frequency reduces the power up to 50{\%} or 30{\%} for low-quality single frame (photo) and sequential frame (video) capturing, respectively; (ii) by entering low-power standby mode between frames, an image sensor achieves almost constant energy per pixel for video capture at low frame rates, resulting in an additional 40{\%} power reduction. We also propose architectural modifications to the image sensor that would further improve operational efficiency. Finally, we use computer vision benchmarks to show the performance and efficiency tradeoffs that can be achieved with existing image sensors. For image registration, a key primitive for image mosaicking and depth estimation, we can achieve a 96{\%} success rate at 3 FPS and 0.1 MP resolution. At these quality metrics, an optimal clock frequency reduces image sensor power consumption by 36{\%} and aggressive standby mode reduces power consumption by 95{\%}.",
keywords = "Computer vision, Energy efficiency, Energy proportionality, Image sensor, Mobile systems",
author = "Robert LiKamWa and Bodhi Priyantha and Matthai Philipose and Lin Zhong and Paramvir Bahl",
year = "2013",
doi = "10.1145/2462456.2464448",
language = "English (US)",
isbn = "9781450316729",
pages = "69--81",
booktitle = "MobiSys 2013 - Proceedings of the 11th Annual International Conference on Mobile Systems, Applications, and Services",

}

TY - GEN

T1 - Energy characterization and optimization of image sensing toward continuous mobile vision

AU - LiKamWa, Robert

AU - Priyantha, Bodhi

AU - Philipose, Matthai

AU - Zhong, Lin

AU - Bahl, Paramvir

PY - 2013

Y1 - 2013

N2 - A major hurdle to frequently performing mobile computer vision tasks is the high power consumption of image sensing. In this work, we report the first publicly known experimental and analytical characterization of CMOS image sensors. We find that modern image sensors are not energy-proportional: energy per pixel is in fact inversely proportional to frame rate and resolution of image capture, and thus image sensor systems fail to provide an important principle of energy-aware system design: trading quality for energy efficiency. We reveal two energy-proportional mechanisms, supported by current image sensors but unused by mobile systems: (i) using an optimal clock frequency reduces the power up to 50% or 30% for low-quality single frame (photo) and sequential frame (video) capturing, respectively; (ii) by entering low-power standby mode between frames, an image sensor achieves almost constant energy per pixel for video capture at low frame rates, resulting in an additional 40% power reduction. We also propose architectural modifications to the image sensor that would further improve operational efficiency. Finally, we use computer vision benchmarks to show the performance and efficiency tradeoffs that can be achieved with existing image sensors. For image registration, a key primitive for image mosaicking and depth estimation, we can achieve a 96% success rate at 3 FPS and 0.1 MP resolution. At these quality metrics, an optimal clock frequency reduces image sensor power consumption by 36% and aggressive standby mode reduces power consumption by 95%.

AB - A major hurdle to frequently performing mobile computer vision tasks is the high power consumption of image sensing. In this work, we report the first publicly known experimental and analytical characterization of CMOS image sensors. We find that modern image sensors are not energy-proportional: energy per pixel is in fact inversely proportional to frame rate and resolution of image capture, and thus image sensor systems fail to provide an important principle of energy-aware system design: trading quality for energy efficiency. We reveal two energy-proportional mechanisms, supported by current image sensors but unused by mobile systems: (i) using an optimal clock frequency reduces the power up to 50% or 30% for low-quality single frame (photo) and sequential frame (video) capturing, respectively; (ii) by entering low-power standby mode between frames, an image sensor achieves almost constant energy per pixel for video capture at low frame rates, resulting in an additional 40% power reduction. We also propose architectural modifications to the image sensor that would further improve operational efficiency. Finally, we use computer vision benchmarks to show the performance and efficiency tradeoffs that can be achieved with existing image sensors. For image registration, a key primitive for image mosaicking and depth estimation, we can achieve a 96% success rate at 3 FPS and 0.1 MP resolution. At these quality metrics, an optimal clock frequency reduces image sensor power consumption by 36% and aggressive standby mode reduces power consumption by 95%.

KW - Computer vision

KW - Energy efficiency

KW - Energy proportionality

KW - Image sensor

KW - Mobile systems

UR - http://www.scopus.com/inward/record.url?scp=84881126474&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84881126474&partnerID=8YFLogxK

U2 - 10.1145/2462456.2464448

DO - 10.1145/2462456.2464448

M3 - Conference contribution

AN - SCOPUS:84881126474

SN - 9781450316729

SP - 69

EP - 81

BT - MobiSys 2013 - Proceedings of the 11th Annual International Conference on Mobile Systems, Applications, and Services

ER -