High-speed min-max bilateral filter-based image dehazing by using GPGPU
Shota Furukawa, Takahiro Fukuda, Takanori Koga, Noriaki Suetake, Eiji Uchino · 2014
Image processing techniques for the removal of haze such as smoke and fog from digital images have been actively researched. In the haze removing from an image, it is necessary to estimate the transmission map and the global atmospheric light in the image. The authors have proposed the estimation method based on the minimum and maximum bilateral filters. However, the computational cost of the filters are very high. The purpose of this study is to accelerate the speed of the proposed method by using the general-purpose computing on graphics processing units (GPGPU) aiming for the proposed method to be actually used in the practical applications. Through the experiments, it is confirmed that the calculation speed of the method is drastically accelerated enough for the practical use.