Two-Dimensional Variational Mode Decomposition and Its Application in the Field Of Road Surface Images Denoising
Download as PDF
DOI: 10.25236/iccpb.2018.017
Author(s)
Zhongbin Wei, Song Zhao, Kun Wu
Corresponding Author
Zhongbin Wei
Abstract
In order to extract road surface features and discriminate the condition of road surface, it is necessary to remove noises and maintain marginal information, based on this, a new non-recursive Two-Dimensional Variational Mode Decomposition (2D-VMD) algorithm is introduced. The algorithm uses non-recursive sifting method to analyse signals in frequency domain, which can inhibit the mode and frequency mixing better, compared to previous some mode decomposition algorithms, such as Bidimensional Empirical Mode Decomposition (BEMD) and Bidimensional ensemble Empirical Mode Decomposition (BEEMD). Thus, it can separate the noises and preserve the original information of images at the greatest extent. By means of contrast experiments, the results indicate that 2D-VMD can remove noises and maintain marginal information effectively with lower Error Rate (2.2%) and higher Power Signal-to-Noise Ratio (29.3012dB). It is feasible to adopt the algorithm in the field of road surface image denoising of road engineering.
Keywords
Road engineering, road surface image denoising, marginal information, Two-Dimensional Variational Mode Decomposition, Bidimensional Empirical Mode Decompositi-on(BEMD), Bidimensional ensemble Empirical Mode Decomposition(BEEMD)