Shin Y S , Cho Y , Pandey G , et al. [IEEE OCEANS 2016 MTS/IEEE Monterey - Monterey, CA, USA (2016.9.19-2016.9.23)] OCEANS 2016 MTS/IEEE Monterey - Estimation of ambient light and transmission map with common convolutional architecture[J]. 2016:1-7.
1. Overview
1.1. Motivation
- the scattering of light from water particles along with the attenuation and change in the color of different wavelengths of ambient light
In this paper, it proposes a method for effective ambient light and transmission estimation in underwater image.
- balance scene recovery model
1.2. Dataset
- synthetic based on ICL-NUIM and SUM database
- 1 million patches
1.3. Model
- multi-scale fusion
- element-wise summation (reduce computation)
- Maxout
1.4. Balance Scene Recovery Model
1.4.1. Haze Model
- can not achieve the recovery of the original scene radiance in an underwater environment
- the attenuation of ambient light in an underwater environment is not only dependent upon the distance travelled and density of particles in the path of the light but also depends on the color/wavelength of the light
- For instance, the light intensity of red channel rapidly decreases whereas green channel decreases slowly
1.4.2. Proposed Model
- A_b. balanced ambient light
- a_b. fixed vector which represents the balanced ambient light inRGB
1.5. Experiments
