A Gradient BYY Harmony Learning Algorithm for Straight Line Detection

  • Authors:
  • Gang Chen;Lei Li;Jinwen Ma

  • Affiliations:
  • Department of Information Science, School of Mathematical Sciences and LAMA, Peking University, Beijing, China 100871;Department of Information Science, School of Mathematical Sciences and LAMA, Peking University, Beijing, China 100871;Department of Information Science, School of Mathematical Sciences and LAMA, Peking University, Beijing, China 100871

  • Venue:
  • ISNN '08 Proceedings of the 5th international symposium on Neural Networks: Advances in Neural Networks
  • Year:
  • 2008

Quantified Score

Hi-index 0.00

Visualization

Abstract

Straight line detection is a basic problem in image processing and has been extensively studied from different aspects, but most of the existing algorithms need to know the number of straight lines in an image in advance. However, the Bayesian Ying-Yang (BYY) harmony learning can make model selection automatically during parameter learning for the Gaussian mixture modeling, which can be further applied to detecting the correct number of straight lines automatically by representing the straight lines with Gaussians or Gaussian functions. In this paper, a gradient BYY harmony learning algorithm is proposed to detect the straight lines automatically from an image as long as the pre-assumed number of straight lines is larger than the true one. It is demonstrated by the simulation and real image experiments that this gradient BYY harmony learning algorithm can not only determine the number of straight lines automatically, but also detect the straight lines accurately against noise.