Improved Dark Channel Prior for Image Defogging Using RGB and YCbCr Color Space

Environmental factors such as fog and haze affect the image quality and make it unsuitable for automated systems, such as intelligent vehicles, surveillance, and outdoor object recognition, which require images with clear visibility for processing and decision making. In general, reconstruction of fog-free image from a single input image is quite challenging. Dark channel prior (DCP) method is used to estimate atmospheric light for the purpose of image defogging. This paper presents a DCP-based image defogging method with improved transmission map to avoid blocking artifacts. The transmission maps are computed for RGB and YCbCr color spaces. Three transmission maps for the R, G, and B channels are utilized to compute a mean transmission map. In the YCbCr color space, Y channel is used to calculate the transmission map. The two transmission maps are refined by preserving edge information for constructing two intermediate images, which are assigned different weights to get the enhanced defogged output. The proposed method is evaluated against the current state-of-the-art approaches, and the experimental results based on structural similarity index, fog effect, anisotropic quality index and degradation score are calculated, which show that the defogged images reconstructed using the proposed method achieved better results with lower fog effect, similarity index, degradation score and higher quality index value. Reconstructed image has better contrast and luminance which is perceptually more appealing to the human visual system.