Summary: | 碩士 === 逢甲大學 === 資訊工程學系 === 105 === The air pollution and foggy weather often result in serious distortion while taking
photos or recognizing patterns. He et al. have introduced the dark channel prior to
solve this dehazing problem. Unfortunately, it cannot function well once the color
difference of target image is large. More precisely, the dehazed result looks unnatural.
Thus, we aim to develop a brand-new visibility dehazing technique based on the
channel-weighted analysis and illumination tuning. The channel-weighted analysis is
adopted to eliminate the unnatural effect, while the illumination tuning is applied to
refine the details. Simulation results have demonstrated that the new method can
guarantee the readability of a hazed image after removing noise, including the foggy
photo and sandstorm one.
|