In computer vision, illumination is considered to be a problem that needs to be ‘solved’.

1. Plain Methods

   1. Histogram Equalization increase contrast of image

      

      Original HE

      

      CLAHE

   2. Gamma Correction carried out on each pixel individually in a non-linear way. Although GC can promote the brightness especially of dark pixels, it does not consider the relationship of a certain pixel with its neighbors.

   3. MinMax Normalization

      

   4. Standardization

      

2. Traditional Illumination-based Methods Strategies in this category are aware of the concept of illumination The key assumption is that the (color) image can be decomposed into two components, i.e. reflectance and illumination.

   1. single-scale Retinex (SSR)
   2. multi-scale Retinex (MSR)
   3. NPE
   4. weighted variational model for simultaneous reflectance and illumination estimation (SRIE)

3. Deep Learning-based Methods

   

   

   http://mc.nankai.edu.cn/ll/

reference

• https://royalsocietypublishing.org/doi/10.1098/rsfs.2018.0008