Twitter LinkedIn
Return to Wellspring

Projects Sharing Researchers

Researchers

Communicate

Dynamically Reconfigurable Architectures for Time Varying Image Constraints (DRASTIC)

Details

Project TitleDynamically Reconfigurable Architectures for Time Varying Image Constraints (DRASTIC)
Track Code2013-028
Short Description

A system and method for improving the resource management in computer systems, which has pareto-optimal realizations (i.e. able to maximize resources) to meet time-varying energy, power, performance, and accuracy constraints. 

Abstract

This system incorporates dynamically reconfigurable signal, image, and video processing architectures, through using multi-objective optimization to determine one or more pareto-optimal realizations for the power, performance, and accuracy space, and selecting among the one or more pareto-optimal realizations to meet time-varying resource constraints of performing the specified task. This scalable and efficient hardware can thus, densely sample the power/energy, bitrate, and image quality space when applied to such digital content applications as 2D/3D filter-banks for image analysis and High-Efficiency Video Coding, and determine appropriate changes to such operating modes as minimum power, maximum accuracy, maximum image quality, minimum bitrate, and maximum throughput. Thus, this innovation offers improvements to existing technology within the broader image processing and computer architecture communities.

 
Tagsdigital video, Resource Management, image processing
 
Posted DateJul 5, 2013

Researcher

Name
Marios Pattichis
Yuebing Jiang
Daniel Llamocca Obregon

Manager

Name
Melissa Castillo

Background

As the demand for processing and communications of digital video content continues to grow, the demands for proper processing of such content on computer systems must also evolve. While a computer system is operating, system resources are spent to manage various aspects of the computer system. Certain systems are configured before operation to prioritize certain objectives of the computer system over others. The outcome is that sacrifices in such aspects as power, energy, performance, or accuracy must be made in order for a computer system to allocate enough resources to one of these, at the expense of the others. What would be more beneficial, especially when dealing with the demands of image processing and digital video content, would be for a system that can meet the time-varying demands of these various system aspects and maximize all to their fullest, while allowing for an appropriate amount of balance.

Technology Description

Researchers at the University of New Mexico have developed a system and method for improving the resource management in computer systems, which has pareto-optimal realizations (i.e. able to maximize resources) to meet time-varying energy, power, performance, and accuracy constraints.  This system incorporates dynamically reconfigurable signal, image, and video processing architectures, through using multi-objective optimization to determine one or more pareto-optimal realizations for the power, performance, and accuracy space, and selecting among the one or more pareto-optimal realizations to meet time-varying resource constraints of performing the specified task. This scalable and efficient hardware can thus, densely sample the power/energy, bitrate, and image quality space when applied to such digital content applications as 2D/3D filter-banks for image analysis and High-Efficiency Video Coding, and determine appropriate changes to such operating modes as minimum power, maximum accuracy, maximum image quality, minimum bitrate, and maximum throughput. Thus, this innovation offers improvements to existing technology within the broader image processing and computer architecture communities.

Advantages/Applications

  • A system which is able to accomplish pareto-optimal realizations/determinations
  • Hardware is scalable and efficient
  • Holds applications in digital video/image processing and computer architecture
  • Maximizes resource allocation and management within computer systems

Publications

INQUIRES

STC has filed intellectual property on this exciting new technology and is currently exploring commercialization options. If you are interested in information about this or other technologies, please contact Arlene Mirabal at amirabal@stc.unm.edu or 505-272-7886.

Files

File Name Description
9,542,198 Issued Patent None Download
9,111,059 Issued Patent None Download

Intellectual Property

Patent Number Issue Date Type Country of Filing
9,542,198 Jan 10, 2017 Continuation United States
9,111,059 Aug 18, 2015 Utility United States