Abstract

Volcanic degassing is often happened on top of Mt Kelud. One of the indicator of its activity is the changing of the lake color. An observation has been conducted for years to watch the condition of the lake using a CCTV camera, the volcanology officer. Since it is located on top of the mountain, weather condition always change from clear, cloudy, or hazy. Obviously, the vision of the camera is limited or even blocked by the haze or fog. In this paper, we use a haze removal technique based on dark channel prior. The problem comes up because the airlight hard to estimate on non-sky area. By comparing different size of the patches, we found that, in our case, for smaller patch the airlight closes to its maximum limit. This airlight almost equals to sky region. For image improvement, we use guided image filter to maintain image quality after transmission map estimation. According to FADE (Fog Aware Density Evaluator) the fog density estimated to be 0.6333 and SSIM (Structural Similarity) 0.9977.

Original languageEnglish
Title of host publication20th International Computer Science and Engineering Conference
Subtitle of host publicationSmart Ubiquitos Computing and Knowledge, ICSEC 2016
PublisherInstitute of Electrical and Electronics Engineers Inc.
ISBN (Electronic)9781509044207
DOIs
Publication statusPublished - 21 Feb 2017
Event20th International Computer Science and Engineering Conference, ICSEC 2016 - Chiang Mai, Thailand
Duration: 14 Dec 201617 Dec 2016

Publication series

Name20th International Computer Science and Engineering Conference: Smart Ubiquitos Computing and Knowledge, ICSEC 2016

Conference

Conference20th International Computer Science and Engineering Conference, ICSEC 2016
Country/TerritoryThailand
CityChiang Mai
Period14/12/1617/12/16

Keywords

  • Airlight
  • Dark channel prior
  • Degassing
  • Fog density
  • Guided image filter
  • Haze removal
  • Lake color
  • Volcano

Fingerprint

Dive into the research topics of 'Visibility restoration of lake crater hazy image based on dark channel prior'. Together they form a unique fingerprint.

Cite this