Vol.28, No.2, 2021, pp.417-428, doi:10.32604/iasc.2021.016802
OPEN ACCESS
ARTICLE
A Fog Covered Object Recognition Algorithm Based On Space And Frequency Network
  • Ying Cui1, Shi Qiu2,*, Tong Li3
1 College of Equipment Management and Support, Engineering University of PAP, Xi’an, 710086, China
2 Key Laboratory of Spectral Imaging Technology CAS, Xi’an Institute of Optics and Precision Mechanics, Chinese Academy of Sciences, Xi’an, 710119, China
3 University of California, Los Angeles, California, 90095, USA
* Corresponding Author: Shi Qiu. Email:
Received 12 January 2021; Accepted 19 February 2021; Issue published 01 April 2021
Abstract
It is difficult to recognize a target object from foggy images. Aiming at solving this problem, a new algorithm is thereby proposed. Fog concentration estimation is the prerequisite for image defogging. Due to the uncertainty of fog distribution, a fog concentration estimation model is accordingly proposed. Establish the brightness and gradient model in the spatial domain, and establish the FFT model in the frequency domain. Also, a multiple branch network framework is established to realize the qualitative estimation of the fog concentration in images based on comprehensive analysis from spatial and frequency domain levels. In the aspect of foggy image target recognition, a residual network is introduced based on Fast RCNN network structure. The fog concentration information is added into the target recognition function to realize accurate recognition of target from foggy images. Experimental results show that the accuracy of AOM is higher than 81 under different fog concentration indoors and outdoors.
Keywords
Foggy image; fog concentration; estimation; target object recognition; fog removal; deep learning
Cite This Article
Y. Cui, S. Qiu and T. Li, "A fog covered object recognition algorithm based on space and frequency network," Intelligent Automation & Soft Computing, vol. 28, no.2, pp. 417–428, 2021.
This work is licensed under a Creative Commons Attribution 4.0 International License , which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.