基于IEA-DWT预处理的弱光环境图像特征匹配研究

郑冬; 冯鹏; 谭周楠; 唐曾; 伍成金; 杜炜

doi:10.11884/HPLPB202638.250349

基于IEA-DWT预处理的弱光环境图像特征匹配研究

doi: 10.11884/HPLPB202638.250349

郑冬^1,,
冯鹏^2, ,,
谭周楠¹,
唐曾¹,
伍成金¹,
杜炜¹

1.
重庆电力高等专科学校信息工程学院，重庆 400053
2.
重庆大学光电工程学院，重庆 400044

基金项目: 重庆市教育委员会科学技术研究项目（KJQN202402606）；重庆市科委技术创新与应用发展专项项目(cstc2021jscx-gksbX0056)；重庆电力高等专科学校教研项目（D-JY202523）

详细信息

作者简介:
郑　冬，1989367133@qq.com

通讯作者:
冯　鹏，coe-fp@cqu.edu.cn。

中图分类号: TP391
计量
- 文章访问数: 47
- HTML全文浏览量: 20
- PDF下载量: 14
- 被引次数: 0
出版历程
- 收稿日期: 2025-10-15
- 修回日期: 2025-02-13
- 录用日期: 2026-01-22
- 网络出版日期: 2026-03-28

Research on image feature matching under low-light environment based on IEA-DWT preprocessing

1.
College Information Engineering , Chongqing Electric Power College, Chongqing 400053, China
2.
College of Optoelectronic Engineering, Chongqing University, Chongqing 400044, China

摘要

摘要: 弱光环境直接导致图像亮度不足、细节湮没且噪声显著，使传统特征匹配算法性能骤降，误匹配率激增。基于此，提出了一种融合光照增强算法（IEA）与离散小波变换（DWT）的新型预处理算法，旨在通过多阶段图像优化为后续特征匹配提供高质量输入。该算法运用IEA自适应提升图像整体亮度与局部对比度，再通过DWT多尺度分解，对分离的低频分量进行主体结构保留，对不同方向的高频分量进行阈值处理，进而实现边缘增强与降噪，最终将优化图像输入SURF算法完成特征提取与匹配。为了验证该算法的有效性，开展了极弱光、微弱光、弱光环境下的图像特征匹配。实验结果表明，相较于单一IEA或DWT预处理方法，所提算法效果最为显著，平均正确特征匹配对达到261，提高6.53倍。新型预处理算法IEA-DWT可以有效解决弱光环境下图像特征匹配中细节模糊和噪声干扰的核心问题，为弱光场景下的特征匹配提供了可靠的技术支撑。
- 弱光环境 /
- 光照增强算法 /
- 离散小波变换 /
- 预处理
Abstract: Background
Low-light environments directly result in insufficient image brightness, obscured details, and significant noise, which lead to a sharp decline in the performance of traditional feature matching algorithms and a surge in the false matching rate.
Purpose
A novel preprocessing algorithm integrating the Illumination Enhancement Algorithm (IEA) and Discrete Wavelet Transform (DWT) is proposed, aiming to provide high-quality input for subsequent feature matching through multi-stage image optimization.
Methods
The algorithm utilizes IEA to adaptively enhance the overall brightness and local contrast of the image. Through DWT multi-scale decomposition, it preserves the main structure of the separated low-frequency components and performs threshold processing on high-frequency components in different directions, thereby achieving edge enhancement and noise reduction. Finally, the optimized image is input into the SURF algorithm to complete feature extraction and matching. To verify the effectiveness of the proposed algorithm, image feature matching experiments were conducted under extremely low-light, faintly low-light, and low-light environments.
Results
The experimental results show that compared with the single IEA or DWT preprocessing methods, the proposed algorithm achieves the most significant effect, with the average number of correct feature matching pairs reaching 261, which is a 6.53-fold increase.
Conclusions
The new preprocessing algorithm IEA-DWT can effectively solve the core problems of blurred details and noise interference in image feature matching in low-light environments, and provides reliable technical support for feature matching in low-light scenes.
- low-light environment /
- illumination enhancement algorithm /
- discrete wavelet transform /
- preprocessing

HTML全文

图 1 弱光环境下图像特征匹配流程图

Figure 1. Flowchart of image feature matching in low-light environment

下载: 全尺寸图片幻灯片

图 2 实验平台搭建示意图

Figure 2. Schematic diagram of experimental platform construction

下载: 全尺寸图片幻灯片

图 3 弱光环境图像数据集

Figure 3. Low-light environment image dataset

下载: 全尺寸图片幻灯片

图 4 IEA-DWT算法初步验证流程图

Figure 4. Preliminary verification flow chart of IEA-DWT algorithm

下载: 全尺寸图片幻灯片

图 5 特征匹配实验结果

Figure 5. Feature matching experiment results

下载: 全尺寸图片幻灯片

图 6 粗迭代图像特征匹配对1

Figure 6. Coarse iterative image feature matching pair 1

下载: 全尺寸图片幻灯片

图 7 粗迭代图像特征匹配对2

Figure 7. Coarse iterative image feature matching pair 2

下载: 全尺寸图片幻灯片

图 8 细迭代图像特征匹配对

Figure 8. Iterative image feature matching pairs

下载: 全尺寸图片幻灯片

图 9 极弱光环境特征匹配算法对比效果图

Figure 9. Comparative effect diagram of extreme low-light environment feature matching algorithms

下载: 全尺寸图片幻灯片

图 10 微弱光环境特征匹配算法对比效果图

Figure 10. Comparative effect diagram of feature matching algorithm for weak light environment

下载: 全尺寸图片幻灯片

图 11 弱光环境特征匹配算法对比效果图

Figure 11. Comparative effect diagram of weak light environment feature matching algorithms

下载: 全尺寸图片幻灯片

图 12 IEA-DWT-SURF算法弱环境灰度分布直方图

Figure 12. IEA-DWT-SURF algorithm grayscale distribution histogram

下载: 全尺寸图片幻灯片

图 13 三种算法正确特征匹配对数目对比图

Figure 13. Comparison chart of the number of correct feature matching pairs for three algorithms

下载: 全尺寸图片幻灯片

表 1 特征匹配参数表

Table 1. Feature match parameter table

algorithm	Classic SURF	IEA-SURF	DWT-SURF	IEA-DWT-SURF
feature matching pairs	90	268	246	306

下载: 导出CSV

表 2 光照增强系数迭代数据表

Table 2. Light enhancement coefficient iteration data table

coefficient/$ \alpha $	matching pairs	coefficient/$ \alpha $	matching pairs	coefficient/$ \alpha $	matching pairs
0	0	2.4	288	5.6	300
0.2	308	2.6	299	5.8	289
0.4	281	2.8	300	6.0	271
0.6	246	3	304	6.2	276
0.8	255	3.2	308	6.4	283
1	60	4.2	305	6.6	279
1.2	300	4.4	310	6.8	293
1.4	274	4.6	293	7	284
1.6	280	4.8	295	7.2	281
1.8	277	5	299	7.4	0
2	280	5.2	305	7.6	0
2.2	284	5.4	299	7.8～8.0	0

下载: 导出CSV

表 3 小波基筛选数据表

Table 3. Wavelet-based filtering data table

wavelet basis	db1	db2	db3	db4	db5	db6	db7	db8	db9	db10
matching pairs	246	245	246	246	246	241	246	248	247	247

下载: 导出CSV

表 4 各算法特征匹配数目

Table 4. Number of matching features of each algorithm

Algorithm	Dataset	Correct match number/permit	Enhancement coefficient	Promote results/double
Classic SURF	test set 1	0	－	－
	test set 2	17	－	－
	test set 3	103	－	－
IEA-SURF	test set 1	48	3.6	48.0
	test set 2	146	3.0	8.6
	test set 3	385	0.2	3.7
DWT-SURF	test set 1	50	－	50.0
	test set 2	145	－	8.5
	test set 3	337	－	3.3
IEA-DWT-SURF	test set 1	76	3.6	76.0
	test set 2	228	3.0	13.4
	test set 3	480	0.2	4.6

下载: 导出CSV

参考文献(22)

[1]	夏立琼, 陈铭, 王鹏, 等. 基于机器视觉方法的诊断设备自动瞄准技术[J]. 强激光与粒子束, 2023, 35: 112002 doi: 10.11884/HPLPB202335.230317 Xia Liqiong, Chen Ming, Wang Peng, et al. Machine vision aided method for the autonomic diagnostic alignments[J]. High Power Laser and Particle Beams, 2023, 35: 112002 doi: 10.11884/HPLPB202335.230317
[2]	向浩鸣, 夏晓华, 葛兆凯, 等. 仿人眼双目图像特征点提取与匹配方法[J]. 哈尔滨工业大学学报, 2024, 56(4): 92-100 Xiang Haoming, Xia Xiaohua, Ge Zhaokai, et al. Feature point extraction and matching method of humanoid-eye binocular images[J]. Journal of Harbin Institute of Technology, 2024, 56(4): 92-100
[3]	黄盼, 何鹏, 杨兴, 等. 基于自适应融合和显微成像的乳腺肿瘤分级网络[J]. 光电工程, 2023, 50: 220158 doi: 10.12086/oee.2023.220158 Huang Pan, He Peng, Yang Xing, et al. Breast tumor grading network based on adaptive fusion and microscopic imaging[J]. Opto-Electronic Engineering, 2023, 50: 220158 doi: 10.12086/oee.2023.220158
[4]	马文浩, 陈颖, 李铖昊, 等. 融合注意力与频域特征的遥感图像配准[J]. 激光杂志, 2025, 46(9): 71-81 doi: 10.14016/j.cnki.jgzz.2025.09.071 Ma Wenhao, Chen Ying, Li Chenghao, et al. Remote sensing image registration integrating attention and frequency domain feature[J]. Laser Journal, 2025, 46(9): 71-81 doi: 10.14016/j.cnki.jgzz.2025.09.071
[5]	汪建伟, 游疆, 万敏, 等. 复杂背景下的低空无人机检测与跟踪算法[J]. 强激光与粒子束, 2023, 35: 079001 doi: 10.11884/HPLPB202335.230026 Wang Jianwei, You Jiang, Wan Min, et al. Low-altitude UAV detection and tracking algorithms in complex backgrounds[J]. High Power Laser and Particle Beams, 2023, 35: 079001 doi: 10.11884/HPLPB202335.230026
[6]	Tang Guoliang, Liu Zhijing, Xiong Jing. Distinctive image features from illumination and scale invariant keypoints[J]. Multimedia Tools and Applications, 2019, 78(16): 23415-23442. doi: 10.1007/s11042-019-7566-8
[7]	Zhang Jianguang, Li Yongxia, Tai An, et al. Motion video recognition in speeded-up robust features tracking[J]. Electronics, 2022, 11: 2959. doi: 10.3390/electronics11182959
[8]	Zhang Shaojie, Wang Yinghui, Ma Jiaxing, et al. Mults-ORB: multistage oriented FAST and rotated brief[J]. Mathematics, 2025, 13: 2189. doi: 10.3390/math13132189
[9]	Dai Yong, Wu Jiaxin. An improved ORB feature extraction algorithm based on enhanced image and truncated adaptive threshold[J]. IEEE Access, 2023, 11: 32073-32081. doi: 10.1109/ACCESS.2023.3261665
[10]	Tang Qingling, Wang Xuanxi, Zhang Meng, et al. Image matching algorithm based on improved AKAZE and Gaussian mixture model[J]. Journal of Electronic Imaging, 2023, 32: 023020. doi: 10.1117/1.jei.32.2.023020
[11]	Al-Ameen Z. Nighttime image enhancement using a new illumination boost algorithm[J]. IET Image Processing, 2019, 13(8): 1314-1320. doi: 10.1049/iet-ipr.2018.6585
[12]	Latif I H, Abdulredha S H, Hassan S K A. Discrete wavelet transform-based image processing: a review[J]. Al-Nahrain Journal of Science, 2024, 27(3): 109-125. doi: 10.22401/ANJS.27.3.13
[13]	Guo Zijun, Wang Chao. Low light image enhancement algorithm based on retinex and dehazing model[C]//Proceedings of the 6th International Conference on Robotics and Artificial Intelligence. 2020: 84-90.
[14]	李冰, 叶猛, 颉卓凡, 等. 基于自适应光照的输电线路自然暗光图像增强方法[J]. 光学学报, 2025, 45: 2310001 doi: 10.3788/AOS251206 Li Bing, Ye Meng, Xie Zhuofan, et al. Adaptive illumination-based natural low-light image enhancement method for transmission lines[J]. Acta Optica Sinica, 2025, 45: 2310001 doi: 10.3788/AOS251206
[15]	Yang Wei, Wang Shuai, Wu Jiaqi, et al. A low-light image enhancement method for personnel safety monitoring in underground coal mines[J]. Complex & Intelligent Systems, 2024, 10(3): 4019-4032. doi: 10.1007/s40747-024-01387-2
[16]	Jin Haiyan, Li Long, Su Haonan, et al. Learn to enhance the low-light image via a multi-exposure generation and fusion method[J]. Journal of Visual Communication and Image Representation, 2024, 100: 104127. doi: 10.1016/j.jvcir.2024.104127
[17]	黄显辉, 田原嫄, 郭海涛. 基于离散小波分解和改进的AGCWD算法的声呐图像对比度增强[J]. 海南热带海洋学院学报, 2025, 32(5): 76-90 doi: 10.13307/j.issn.2096-3122.2025.05.09 Huang Xianhui, Tian Yuanyuan, Guo Haitao. Contrast enhancement of sonar images based on discrete wavelet decomposition and improved AGCWD algorithm[J]. Journal of Hainan Tropical Ocean University, 2025, 32(5): 76-90 doi: 10.13307/j.issn.2096-3122.2025.05.09
[18]	Yang Ruoxi, Chen Long, Zhang Ling, et al. Image enhancement via special functions and its application for near infrared imaging[J]. Global Challenges, 2023, 7: 2200179. doi: 10.1002/gch2.202200179
[19]	秦蒙, 郑冬, 冯鹏, 等. 一种预处理OWLCM算法下的图像特征匹配研究方法[J]. 重庆理工大学学报(自然科学), 2025, 39(1): 117-124 doi: 10.3969/j.issn.1674-8425(z).2025.01.015 Qin Meng, Zheng Dong, Feng Peng, et al. A pre-processing OWLCM algorithm for image feature matching research[J]. Journal of Chongqing University of Technology (Natural Science), 2025, 39(1): 117-124 doi: 10.3969/j.issn.1674-8425(z).2025.01.015
[20]	郭昕刚, 许连杰, 程超, 等. 加权核范数最小化和改进小波阈值函数的图像去噪算法[J]. 国防科技大学学报, 2024, 46(2): 238-246 doi: 10.11887/j.cn.202402024 Guo Xingang, Xu Lianjie, Cheng Chao, et al. Image denoising algorithm based on weighted kernel norm minimization and improved wavelet threshold function[J]. Journal of National University of Defense Technology, 2024, 46(2): 238-246 doi: 10.11887/j.cn.202402024
[21]	Shwetar Y J, Haendel M A. Multidimensional quantification of macular cone activity in pattern electroretinography using discrete wavelet transform[J]. Translational Vision Science & Technology, 2025, 14: 17. doi: 10.1167/tvst.14.9.17
[22]	张军华, 胡逸甫, 于正军, 等. 基于图像处理的河流相储层边缘特征增强方法研究[J]. CT理论与应用研究, 2023, 32(4): 450-460 doi: 10.15953/j.ctta.2022.174 Zhang Junhua, Hu Yifu, Yu Zhengjun, et al. Research on the edge feature enhancement of fluvial reservoirs based on image processing[J]. CT Theory and Applications, 2023, 32(4): 450-460 doi: 10.15953/j.ctta.2022.174