Universitas Darussalam Gontor

gunung kelud

[Single Kelud Volcano Lake Crater Image Dehazing Using Color Attenuation Prior and Adaptive Gamma Correction]

Visibility of outdoor images captured in bad weather often decreases due to fog, sandstorms, and so on. Poor visibility, caused by atmospheric phenomena, is a factor in the failure of computer vision applications, such as external object recognition systems, obstacle detection systems or video surveillance systems. Due to the last eruption, CCTV cameras have been installed on top of Mt. Kelud summit to observe the crater of the lake and its surroundings.

However, the observation camera experienced interference due to the fog. Not only that, the removal of fog from an image with complicated structures, halo effect, and color distortion is challenging image recovery techniques. This study aims to reduce the fog and improve the visibility of the foggy image. In this article, a new dehazing method is proposed that combines the Color Attenuation Prior (CAP) and Adaptive Gamma Correction (AGC) methods.

This is divided into three main modules, namely the depth estimation module (DispE), the transmission map enhancement module (TME), and the restoration module (ImRec). The proposed DispE module utilizes depth estimation techniques from CAP. While the TME module adopts the AGC technique. Thus, the halo effect on the image can be avoided and the estimation of an effective transmission map can be achieved. Furthermore, the ImRec module uses a transmission map output from TME to correct the color distortion of the crater image. Experimental results show that the proposed method can reduce haze without causing halo and color distortion effects. Subsequent research focused on machine learning based methods.

Keywords: adaptive gamma correction, color attenuation prior, dark channel prior, dehazing, haze.

Oddy Virgantara Putra, Aziz Musthafa

Full Text:


Taufiq Affandi

Taufiq Affandi

Leave a Reply

Your email address will not be published. Required fields are marked *