Comparative evaluation of noise insensitivity of linear edge detection techniques
M. Yu. Khomyakov · Pattern Recognition and Image Analysis · 2011
This work presents a comparative estimate of white-noise insensitivity of the most requested linear edge detection techniques dealing with grayscale images. We consider detectors based on the first and second derivatives of the image brightness. The comparative evaluation was based on statistical analysis. The Kirsh and Frei-Chen operators were found to be less sensitive to noise.