6 August 2009 Wavelet edge detection based on self-adjusted directional derivative
Author Affiliations +
Abstract
A multi-scale wavelet edge detection algorithm based on directional derivative which can be self-adjusted is proposed. The high precision and the excellent immunity from noise are achieved. The standard methods of wavelet transform images along horizontal and vertical directions and suit the detection of horizontal or vertical edges. If they are used to detect slanting edges, the precision will decline. Other existing wavelet algorithms considering direction information can only process images along some specified directions. The difficulty confronted by these methods is the dilemma between the calculational complexity and the orientation accuracy. In this paper, the approach of wavelet edge detection based on directional derivative which can self-adopt orientation according to edge direction is investigated. The wavelet transforms are carried out on three scales. At each point of an image, the directional derivative is designed locally based on the computational results of the neighboring scale so as to acquire self-adjusting characteristic. This has the advantage to improve precision, and almost not increase the complexity. Besides, the relationship between the Lipschitz exponent and the magnitudes of wavelet transformation is used to restrain noise. Finally the edge detection experiments for noise-stained images were done. The results show that our method can achieve both good visual quality and high PSNR which is enhanced by 3.6 and 6.6 percent respectively comparing with two other wavelet algorithms.
© (2009) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Jun-fang Wu, Jun-fang Wu, Gui-xiong Liu, Gui-xiong Liu, } "Wavelet edge detection based on self-adjusted directional derivative", Proc. SPIE 7384, International Symposium on Photoelectronic Detection and Imaging 2009: Advances in Imaging Detectors and Applications, 73840J (6 August 2009); doi: 10.1117/12.834838; https://doi.org/10.1117/12.834838
PROCEEDINGS
7 PAGES


SHARE
Back to Top