Use of color information for keypoints detection and descriptors construction

  • Authors:

  • Andrey S. Krylov;Dmitry V. Sorokin;Dmitry V. Yurin;Ekaterina V. Semeikina

  • Affiliations:

  • Laboratory of Mathematical Methods of Image Processing, Faculty of Computational Mathematics and Cybernetics, Lomonosov Moscow State University, Moscow, Russia;Laboratory of Mathematical Methods of Image Processing, Faculty of Computational Mathematics and Cybernetics, Lomonosov Moscow State University, Moscow, Russia;Laboratory of Mathematical Methods of Image Processing, Faculty of Computational Mathematics and Cybernetics, Lomonosov Moscow State University, Moscow, Russia;Laboratory of Mathematical Methods of Image Processing, Faculty of Computational Mathematics and Cybernetics, Lomonosov Moscow State University, Moscow, Russia

  • Venue:
  • IScIDE'11 Proceedings of the Second Sino-foreign-interchange conference on Intelligent Science and Intelligent Data Engineering
  • Year:
  • 2011

Quantified Score

Hi-index 0.00

Visualization

Abstract

An effective use of color information in the keypoints detection and descriptors construction is an open task. In the first part of the paper a novel scale-space color blob detection technique is introduced. The effectiveness of proposed technique for keypoints detection is demonstrated in comparison with grayscale detector and alternative color detector. We illustrate the importance of using color keypoints detection technique in the cases when some image features become indistinguishable in grayscale. The most important application of color blob detection scheme is to guarantee operability in the worst case. The importance of color information in descriptors construction process is demonstrated comparing the color and grayscale versions of Gauss-Laguerre keypoints descriptors.The task of matching the images connected by homography transformation was chosen to compare the quality of algorithms. It was found that the use of color information for keypoints descriptors construction for both color and grayscale keypoints detection usually enhance the matching quality. For some cases the use of color information in keypoints detection procedure further improves matching results.