VIDEO ENHANCEMENT USING DEEP LEARNING

Abstract
In recent years, technologies have evolved from simple mobile phones to complicated surveillance monitoring systems capable of capturing and processing video clips. During the video acquisition process, the recorded quality degrades, which is inevitable. Poor illumination and the wrong aperture or shutter speed settings are to blame. This constraint frequently results in photographs with poor quality or images with low contrast and a noisy backdrop. In addition, a video low contrast might be caused by a faulty imaging instrument or a lack of knowledge on the part of the operator. As a result, the available dynamic range is underutilised during video acquisition. As a result, the video finer details are obscured, and the image may appear washed out or strange. Contrast enhancement techniques, which enhance the image visual quality, help to mitigate these issues. The current work aims to address the issues outlined above by employing two distinct approaches. Video compression and contrast enhancement are two different techniques that can be used in conjunction with each other. To increase the quality of videos, the Deep Learning-based Adaptive Cumulative Distribution Based Histogram Enhancement (DLACDHE) technique is applied. The video frames can be more effectively analysed using this hybrid technique. To further reduce noise, the NonDivisional Median Filter is used. When analysing the sounds, the concept of neighbourhood similarity is employed. Using the proposed DLACDHE approach, the study found that it outperforms other methods. In light of the findings, we may conclude that the proposed strategy provides superior contrast enhancement to the already used approaches.

Authors
J. Jasmine
Sri Shakthi Institute of Engineering and Technology, India

Keywords
Automatic Fish Detection, Fish Classification, Fish Species Recognition, Fish Database, Feature Extraction
Published By :
ICTACT
Published In :
ICTACT Journal on Image and Video Processing
( Volume: 12 , Issue: 4 )
Date of Publication :
May 2022
DOI :

Creative Commons License
This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.