基于优势特征融合的核电站水下图像增强  

Underwater Image Enhancement of Nuclear Power Plants Based on Dominant Feature Fusion

作　　者：李佳轩 程竹明 黄三傲 吕天明 王培珍 LI Jiaxuan;CHENG Zhuming;HUANG Sanao;LYU Tianming;WANG Peizhen(School of Electrical&Information Engineering,Anhui University of Technology,Maanshan 243032,China;CGNPC Testing Technology Co.,Ltd,Suzhou 215000,China)

机构地区：[1]安徽工业大学电气与信息工程学院,安徽马鞍山243032 [2]中广核检测技术有限公司,江苏苏州215000

出　　处：《安徽工业大学学报(自然科学版)》2025年第2期169-177,共9页Journal of Anhui University of Technology(Natural Science)

基　　金：安徽省高校自然科学基金重点项目(KJ2019A0085)。

摘　　要：针对核电站水下环境导致图像质量下降如颜色偏差、对比度不足和细节模糊等问题,提出1种基于优势特征融合的核电站水下图像增强方法。在利用自动颜色均衡算法实现图像颜色校正的基础上,分别通过改进的非锐化掩膜算法和加权自适应伽玛校正算法增强图像的锐度和对比度;利用权重图对锐度和对比度增强的图像进行优势特征多尺度融合。以核电站水下原始图像数据集为样本,采用本文方法与其他5种水下图像处理方法进行对比实验,验证本文方法的有效性。结果表明:本文方法能有效解决核电站水下图像颜色偏差、对比度不足、细节模糊等问题,其中水下图像质量评价指标(UIQM)和信息熵的均值总体较高,分别为3.1037,7.5027,与原始图像相比提高幅度分别达121%和9.66%;此外,利用本文方法增强的图像显著增加了特征点匹配对的数目,从而大大提高了视觉特征提取和特征匹配的效率。本文研究可为核电站水下图像分析和设备缺陷智能检测提供新的技术支撑,有助于促进核能的高效利用和可持续发展。In response to the degradation of image quality caused by underwater environments in nuclear power plants,such as color deviation,insufficient contrast,and blurred details,a novel underwater image enhancement method based on the fusion of dominant features was proposed.Based on the implementation of image color correction using the automatic color equalization algorithm,the sharpness and contrast of the images were enhanced through an improved unsharp masking algorithm and a weighted adaptive gamma correction algorithm,respectively.The enhanced images were subjected to multi-scale fusion of dominant features using weight maps.Using a dataset of original underwater images from nuclear power plants as samples,the proposed method was compared with five other underwater image processing techniques to validate its effectiveness.The results demonstrate that the proposed method effectively addresses the issues such as color deviation,insufficient contrast,and blurred details in underwater images of nuclear power plants.The mean values of the underwater image quality measure(UIQM)and information entropy are significantly higher at 3.1037 and 7.5027,respectively,representing improvements of 121%and 9.66%compared to the original images.Furthermore,the enhanced images generated by the proposed method significantly increase in the number of feature point matching pairs,thereby greatly improving the efficiency of visual feature extraction and matching.This research provides new technical support for underwater image analysis and intelligent defect detection of equipment in nuclear power plants,contributing to the efficient utilization and sustainable development of nuclear energy.

关 键 词：核电站 水下图像 颜色校正 伽玛校正 非锐化掩膜 优势特征 多尺度融合 智能检测 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]