Twitter LinkedIn
Return to Wellspring

Projects Sharing Researchers

Researchers

Communicate

The Fast Discrete Periodic Radon Transform for Prime Sized Images: Algorithm, Architecture, and FPGA Implementation

Details

Project TitleThe Fast Discrete Periodic Radon Transform for Prime Sized Images: Algorithm, Architecture, and FPGA Implementation
Track Code2014-068
Short Description

A scalable hardware architecture and associated algorithm for faster computing of the Discrete Periodic Radon Transform (DPRT) for prime-sized images.

Abstract

This research group has come up with a parameterizable approach that can be implemented with limited hardware resources while allowing for the maximum possible execution speed, as measured in the required number of clock cycles. The system is currently under development for computing 2D convolutions.

 
Tagsalgorithm, 2D convolutions, image processing
 
Posted DateJun 11, 2014

Researcher

Name
Marios Pattichis
Daniel Llamocca Obregon
Cesar Carranza

Manager

Name
Melissa Castillo

Background

The Discrete Periodic Radon Transform (DPRT) is an essential component of a wide range of applications in image processing. The DPRT has many important applications that are associated with reconstructing objects from projections, image denoising, image restoration, encryption, and the fast computation of 2D convolutions. An effective method for computing 2D convolutions with limited resources is currently under development at the University of New Mexico. There is a strong market for the effective computation of 2D convolutions since such a technology can be applied to all areas of image processing.

Technology Description

Researchers at the University of New Mexico have developed a scalable hardware architecture and associated algorithm for faster computing of the Discrete Periodic Radon Transform (DPRT) for prime-sized images. This research group has come up with a parameterizable approach that can be implemented with limited hardware resources while allowing for the maximum possible execution speed, as measured in the required number of clock cycles. The system is currently under development for computing 2D convolutions.

Advantages/Applications

  • Balance hardware resources versus execution time
  • Reduces the required hardware resources so that the architecture will fit in smaller devices
  • Larger devices can provide faster implementations
  • Architecture can easily be modified to support the inverse DPRT for prime numbers
  • Architecture is scalable for higher image sizes
  • DPRT for a 509 x 509 image can be computed in 1.31ms
  • Suitable for very-large-scale integration (VLSI) implementations

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
US/2016/0314603 Published Patent Application None Download