Energy and Memory Efficient Analog to Digital Image Processing Method for use in Data Intensive Systems

Case ID:
UA14-046
Invention:

This invention starts from the mixed signal data collection and integrates the sampling scheme with memory and digitization. This integrated interface model will reduce the data amount at the very beginning of data acquisition. Thus, data received at the CPU, as well as power consumption in the CPU, will be greatly reduced. In this implementation, there is developed a level-crossing sampling approach to replace Nyquist sampling. This involves a design of a new front end circuit that combines the level-crossing concept with random selection matrices. The new circuit performs digitization only when there is enough variation in the input and when the random selection matrix chooses this input. Lastly it implements a new sampling concept as an in-memory design to replace the existing Analog to Digital converter (A/D) and data compression mechanisms. 
 

Background:
While most of the existing ideas in low power very large-scale integration (VLSI) design are valid, there is also a need to focus on how we sample the external objects, extract the features, and process the data. This invention provides a completely different approach to low power design.

 

Applications:

  • Improve power and memory efficiency of analog to digital image sensing across a broad range of applications


Advantages:

  • University of Arizona simulation results show orders of magnitude in improvement, as compared to traditional methods
  • Real results from phantom data closely replicate the performance of simulations
  • Combines the level-crossing concept with random selection matrices and can estimate very accurate field inhomogeneity maps over an arbitrary range of values
  • New circuit performs digitization only when there is enough variation in the input and when the random selection matrix chooses this input
  • New sampling concept as an in-memory design to replace the existing Analog to Digital converter (A/D) and data compression mechanisms  
  • Greatly reduces the needed vector pixel comparison storage for image sensing


Status: issued U.S. patent #10,013,384

Patent Information:
Contact For More Information:
Tariq Ahmed
Sr Licensing Manager, College of Engineering
The University of Arizona
tariqa@tla.arizona.edu
Lead Inventor(s):
Janet Roveda
Linda Powers
Keywords: