Enhancing Low Light Image Classification Using MADPIP Approach

Many researchers are interested in finding solutions for face recognition against low light illumination, many succeeded using Histogram Equalization, Retinex Theory, Illumination map estimation model, Frequency domain and Deep learning methods. Nevertheless, not all these mentioned methods can improve all types of face detection and face recognition challenges. Many researchers of face recognition agree and believe that a better-quality input image used for face detection, feature extraction, feature selection, training and testing will produce better detection and recognition. Thus, this paper will evaluate other established and published algorithms and our proposed method called Multiple Adaptive Derivative Passive Image Processing (MADPIP). Our solution does not use only one method but multiple and does not have a single output but multiple outputs. When an image goes through multiple adaptive changes multiple times, it is no longer appropriate to call it an output but rather a multiple-processed output or a multiple adaptive derivative.