图像融合
人工智能
融合
亮度
计算机科学
图像(数学)
计算机视觉
图像增强
对比度增强
模式识别(心理学)
光学
医学
哲学
语言学
物理
磁共振成像
放射科
作者
Yongcheng Han,Wenwen Zhang,Weiji He
出处
期刊:Journal of physics
[IOP Publishing]
日期:2023-06-01
卷期号:2478 (6): 062022-062022
标识
DOI:10.1088/1742-6596/2478/6/062022
摘要
Abstract We propose an efficient and novel framework for low-light image enhancement, which aims to reveal information hidden in the darkness and improve overall brightness and local contrast. Inspired by exposure fusion technique, we employ simulated multi-exposure images fusion to derive bright, natural and satisfactory results, while images are taken under poor conditions such as insufficient or uneven illumination, back-lit and limited exposure time. Specifically, we first design a novel method to generate synthesized images with varying exposure time from a single image. Thus, each image of these artificial sequences contains necessary information for the final desired enhanced result. We then introduce a flexible multi-exposure fusion framework to achieve fused images, which comprises a weight map prediction module and a multi-scale fusion module. Extensive experiments show that our approach can achieve similar or better performance compared to serval state-of-the-art methods.
科研通智能强力驱动
Strongly Powered by AbleSci AI