Low-light image enhancement based on machine learning

This project focuses on low-light image enhancement using machine learning techniques. Images captured in low-light conditions often suffer from low contrast, poor visibility, and unexpected noise, leading to unpleasant subjective feelings and hindering the performance of computer vision tasks like...

Full description

Saved in:
Bibliographic Details
Main Author: Li, Siyang
Other Authors: Jiang Xudong
Format: Final Year Project
Language:English
Published: Nanyang Technological University 2024
Subjects:
Online Access:https://hdl.handle.net/10356/176732
Tags: Add Tag
No Tags, Be the first to tag this record!
Institution: Nanyang Technological University
Language: English
Description
Summary:This project focuses on low-light image enhancement using machine learning techniques. Images captured in low-light conditions often suffer from low contrast, poor visibility, and unexpected noise, leading to unpleasant subjective feelings and hindering the performance of computer vision tasks like object detection and scene understanding. To address these issues, this project explores methods that utilize deep neural networks and classical nonlinear point processing techniques to enhance low-light images. The Zero DCE-Net model is used for training and evaluation. The findings demonstrate the model's efficiency in addressing the challenges of traditional techniques and its potential to improve various computer vision applications. The successful development and evaluation of Zero DCE-Net open avenues for future research and applications in real-world scenarios.