Automatic edge detection using vector distance and partial normalization

  • Authors:
  • Shuhan Chen;Weiren Shi;Kai Wang

  • Affiliations:
  • College of Automation, Chongqing University, Chongqing, People's Republic of China;College of Automation, Chongqing University, Chongqing, People's Republic of China;College of Automation, Chongqing University, Chongqing, People's Republic of China

  • Venue:
  • WSEAS Transactions on Computers
  • Year:
  • 2011

Quantified Score

Hi-index 0.00

Visualization

Abstract

This paper proposes a novel edge detection method for both gray level images and color images, and which can overcome the limitations of gradient-based edge detection methods. A vector distance between feature vector and minimum vector which determines the edge intensity is defined based on four directional summed magnitude differences in a mask, and partial normalization is applied to facilitate threshold selecting. This paper also proposes an improved approach to determine the edge direction. According to the improved edge direction, non-maxima suppression is applied to thin edges, and final edges are extracted automatically using OTSU, even in a changing environment. Extensive experimental results have demonstrated that the proposed method does well in keeping low-contrast edges, selecting threshold and processing time.