A Method of Automatic Detection of Fog Image Based on SVM Classification

Content Provider	Semantic Scholar
Author	Shi, Hesheng
Copyright Year	2017
Abstract	The prerequisite of intelligent image de-fogging is to detect whether fog exists in the image. In this paper, a method to detect fog image from outdoor natural images is proposed. The method normalizes the RGB components and brightness of the image to acquire the minimal correlation between the image intensity and the external light intensity. Then, the two-dimensional discrete Fourier transform is used to obtain the spectrum signature of the image. Together with the gray-level co-occurrence matrix of the image, the spectrum signature is used as a classification feature. The author collects a large number of outdoor fog-free colored images, carries out fogging simulation, and obtains the vector feature library of foggy images through training and evaluation. The author also conducts an experiment to detect foggy and fogless natural images. The experiment proves that the said vector feature library is credible and the proposed method has good ability of detecting fog image, which provides a good precondition and feasible method for intelligent de-fogging.
File Format	PDF HTM / HTML
Volume Number	31
Alternate Webpage(s)	http://revistadelafacultaddeingenieria.com/index.php/ingenieria/article/download/1141/1143
Language	English
Access Restriction	Open
Content Type	Text
Resource Type	Article