Summary: | 碩士 === 國立雲林科技大學 === 電子工程系 === 102 === As technology advances,it is an urgent matter how to make surveillance cameras running in all weather conditions.Because surveillance cameras captured images are often affected by the weather conditions in the real-word.It will miss some important information. And in fact, the current outdoor surveillance system needs defogging function, but most surveillance system use polarized lenses which are not only costly and can not be popular to achieve the effect of defogging on the current market.The existing image fog detection and removal technology can not accurately detect and removal the fog at night environment. Therefore, we propose a novel image fog detection and removal algorithm that uses Modified dark channel prior to detection fog more accurate than existing methods.Next, we distinguish the sky region and object region by texture features, and use different weights to restore image.Thus, we can improve the performance for fog detection and removal.
The experimental results show that the proposed algorithm can increase 28.5% computing speed, improve about 31% image contrast, improve about 16.8% image sharpness, enhance about 6% structural similarity and reduced about 4% over-saturated Pixel number when comparing with the existing methods.In addition, we also ported the fog detection and removal algorithms to embedded systems.And we will also use the embedded system control multiple color temperature LED lights.
|