Multiresolution edge detection based on modified bubble function and SWT

Abstract

A new way based on a modified bubble function and stationary wavelet transform (SWT) is proposed to solve the problem that the conventional edge detection algorithms are sensitive to the noises. Firstly, the traditional bubble function is modified in order to get different time-frequency domain responses and to get filtering effects through adjusting the parameters. Secondly, the modified bubble function is combined with SWT to construct a multiresolution network. By using the modified bubble function to enhance the edges and by using SWT to reduce the noises, the edges can be extracted accurately, effectively and quickly with lower noise. Finally, the experimental results of the proposed edge detection algorithm are verified to be better than that with the traditional bubble function.