An Improved Multi-Scale Fusion Dark Channel Prior Image Dehazing Algorithm
In order to solve the problems of edge blurring,halo and oversaturation of highlighted areas in the dark channel prior image defogging algorithm,an improved dark channel prior image defogging algorithm based on multi-scale fusion is proposed.This algorithm estimates the image edge pixel by pixel by detecting the image,and estimates the non-edge area block by block,so as to weaken the artifacts of the defog image edge.At the same time,the method of combining the dark channel prior and the color attenuation prior is used to give three different attenuation coefficients to the RGB channel to calculate the foreground transmittance estimation map obtained by the dark channel prior.Finally,the fog free image fused with the sky area and the foreground area using the Sigmoid function is restored through the atmospheric scattering model.The experimental results show that the edge artifact has been significantly reduced,and the halo in the highlighted area has also been sig-nificantly improved.Compared with other experiments,the proposed algorithm shows higher efficiency and ap-plicability.
image dehazingmulti-scale fusiondark channel priorcolor attenuation prioratmospheric scattering model
NETL
NSTL
万方数据
