Fast implementation of curve detection based on RHT on dual C6416 system

Guo Man Liu*, Mei Guo Gao, Wen Hua Yang, Tao Pan

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

Abstract

In the application which requests high real-time performance, there still exists a problem, that is, the calculation speed is not fast enough in randomized Hough transform. For the randomized Hough transform algorithm of conic detection and based on its characteristic, a method of its fast implementation in a parallel C6416 system which core calculation units are two high performance DSP chips TMS320C6416 is presented. The core part of this system-digital signal processing board has high performance and simple and symmetrical structure, so that it is easy to implement randomized Hough transform algorithm. With the parallelism of the routine taken into full consideration, which condition is a 800 × 600 gray image, 1000 times randomized. The whole detection process is about 80 ms.

Original languageEnglish
Pages (from-to)1138-1141
Number of pages4
JournalXi Tong Gong Cheng Yu Dian Zi Ji Shu/Systems Engineering and Electronics
Volume27
Issue number6
Publication statusPublished - Jun 2005

Keywords

  • C6416
  • Curve detection
  • Parallel system
  • Randomized Hough transform

Fingerprint

Dive into the research topics of 'Fast implementation of curve detection based on RHT on dual C6416 system'. Together they form a unique fingerprint.

Cite this