Fog removal and enhancement method for UAV aerial images based on dark channel prior

Abstract

The existing UAV aerial image de-fog methods have low image contrast after de-fog, the difference between light and dark image is not obvious, leading to poor de-fog effect. Therefore, an aerial image de-fog enhancement method based on dark channel a priori is proposed. The image variance and absolute gradient mean are combined to get the weight coefficients, and the edge pixels are smoothed by using the multiple decomposition form. The image intensity is calculated and the noise is reduced. A convolution neural network is introduced to calculate the atmospheric transmittance in haze. Based on this, dark channel prior algorithm is used to enhance the light and shade difference of aerial photography image and realise the de-fog enhancement of aerial photography image. To verify the performance of the proposed method, simulation experiments are designed which were compared with the existing methods results in better fog-removing effect, higher contrast and shorter time.

KEYWORDS:

• Dark channel prior
• unmanned aerial vehicle (UAV)
• aerial image
• fog enhancement
• halo artefact
• image denoising

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

The research is supported by: The National Key Research and Development Program of China [grant numbers 2020YFC2004003 and 2020YFC2004002].

Notes on contributors

Fei Xia

Fei Xia, October 1981, Male, Senior Engineer; From September 2000 to July 2004, he graduated from Nanjing University of Posts and Telecommunications with a bachelor’s degree in communication engineering; 2004.9–2007.7 graduated from Nanjing University of Posts and Telecommunications, majoring in communication and information system, with a master’s degree; Now he works in State Grid Jiangsu Information & Telecommunication Company as the deputy director of the information operation inspection centre; He has long been engaged in the research on cloud computing and power information transformation construction. At present; he has published more than 10 papers and participated in 6 provincial company science and technology projects.

Hu Song

Hu Song, October 1986, Male, Senior Engineer; From September 2004 to July 2008, he graduated from Hefei University of Technology with a bachelor’s degree in computer science and technology; From September 2008 to July 2013, he graduated from the University of Science and Technology of China, majoring in computer system structure and obtained a doctorate; Now he works in State Grid Jiangsu Information & Telecommunication Company as a full-time post; He has been engaged in the research of cloud computing and artificial intelligence for a long time; At present, he has published more than 10 papers and participated in 4 provincial company science and technology projects.

Haoxiang Dou

Haoxiang Dou, October 1988, Male, Engineer; From September 2007 to June 2011, he graduated from North China Electric Power University (Baoding), majoring in electrical engineering and automation, with a bachelor’s degree; From September 2011 to April 2014, he graduated from North China Electric Power University with a master’s degree in electronic and communication engineering; Now he works in State grid JIANGSU Information & Telecommunication Company as a full-time post; He has been engaged in the research on digital transformation of power system for a long time; At present, he has published six papers and participated in two provincial science and technology projects.