AN IMPROVISED VIDEO ENHANCEMENT USING MACHINE LEARNING

Abstract

The visual quality of underexposed videos can be increased with the help of a process known as video enhancement. Within the scope of this research, we present a novel approach to enhancing the exposure of movies that are underexposed. Because it has a low barrier to entry theoretically and can reliably give perceptually pleasant outcomes without adding artifacts, it is ideal for a wide variety of practical applications. This is because it is useful for a wide variety of practical applications. We demonstrate the usefulness of the method by displaying improved films of good quality that were made from a variety of different sorts of underexposed videos. A novel approach to the enhancement and editing of video is presented by our method. Rather than relying on a single heuristic transform function, it makes use of human visual perception to adaptively customize the overall visual appearance of the finished product. We believe that our work has the potential to make a big influence in the field of perception-aware video editing and the applications that are related to it, and that it is an important contribution to the community that works on improving videos. The challenge of video enhancement can be formulated as follows: given a video of low quality as the input, produce a video of high quality as the output, but only for specific uses. Whether it be in terms of the video objective clarity or its more subjective qualities, we are interested in learning how we might improve its overall quality.

Keywords

• visual appearance
• video enhancement