传统的Retinex算法针对海上低照度船舶图像增强存在图像过度曝光、细节丢失以及边缘模糊等现象。因此，提出一种结合多尺度Retinex和同态滤波的改进算法。首先，对输入图像的RGB三通道利用改进的单参数同态滤波器进行处理，将处理后的图像转为HSV空间，其中采取自适应增强S分量；由多尺度Retinex算法估计其光照分量并由CLAHE算法均衡化，随之融合Gamma变换处理后的反射分量，融合后通过拉普拉斯算子对其边缘进行增强。最后有效融合校正后的光照分量与边缘增强后的反射分量，输出增强图像。通过实验结果对比分析，证实提出的算法有利于海上低照度船舶图像清晰度的提升及避免细节丢失和图像过曝问题的产生。
The traditional Retinex algorithm for image enhancement of low illuminance ships at sea has some problems, such as overexposure, detail loss and edge blur. Therefore, an improved algorithm combining multiscale Retinex and homomorphic filtering is proposed. First, the RGB three channels of the input image are processed by an improved single parameter homomorphic filter, and the processed image is transformed into HSV space, in which the s-component is adaptively enhanced. The illumination component is estimated by the multi-scale Retinex algorithm and equalized by the CLAHE algorithm, and then the reflection component after gamma transformation is fused. After fusion, its edge is enhanced by Laplacian operator. Finally, the corrected illumination component and the edge enhanced reflection component are effectively fused to output an enhanced image. The comparative analysis of the experimental results proves that the proposed algorithm is conducive to the improvement of the clarity of the marine low illuminance ship image and the generation of detail loss and image overexposure problems.
2024,46(1): 152-157 收稿日期：2022-11-13
DOI：10.3404/j.issn.1672-7649.2024.01.026
分类号：TP391
基金项目：国家级大学生创新创业训练计划项目（202110340036，202210340064）
作者简介：饶伟(1997-),男,硕士研究生,研究方向为图像识别和目标检测
参考文献：
[1] 刘寿鑫, 龙伟, 李炎炎, 等. 基于HSV色彩空间的低照度图像增强[J]. 计算机工程与设计, 2021, 42(9): 2552-2560.
LIU S X, LONG W, LI Y Y, et al. Low illumination image enhancement based on HSV color space[J]. Computer Engineering and Design, 2021, 42(9): 2552-2560.
[2] 田江丽, 李攀. 低照度的船舶图像增强研究[J]. 舰船科学技术, 2021, 43(2): 88-90.
TIAN J L, LI P. Research on ship image enhancement under low illuminance[J]. Ship Science and Technology, 2021, 43(2): 88-90.
[3] 王智奇, 李荣冰, 刘建业, 等. 基于同态滤波和直方图均衡化的图像增强算法[J]. 电子测量技术, 2020, 43(24): 75-80.
WANG Z Q, LI R B, LIU J Y, et al. Image enhancement algorithm based on homomorphic filtering and histogram equalization[J]. Electronic Measurement Technology, 2020, 43(24): 75-80.
[4] 程新. 基于同态滤波的图像增强算法研究[D]. 西安: 西安邮电大学, 2016.
[5] 闫保中, 韩旭东, 何伟. 基于Retinex理论改进的低照度图像增强算法[J]. 应用科技, 2020, 47(5): 74-78.
YAN B Z, HAN X D, HE W. An improved low illumination image enhancement algorithm based on Retinex theory[J]. Applied Science and Technology, 2020, 47(5): 74-78.
[6] 张江鑫, 杨惠. 基于同态高低通滤波与多尺度Retinex的低照度彩色图像增强[J]. 计算机应用与软件, 2021, 38(01): 232-237.
ZHANG J X, YANG H. Low illumination color image enhancement based on homomorphic low-pass filtering and multiscale Retinex[J]. Computer Applications and Software, 2021, 38(01): 232-237.
[7] 李红, 王瑞尧, 耿则勋, 等. 基于多尺度梯度域引导滤波的低照度图像增强算法[J]. 计算机应用, 2019, 39(10): 3046-3052.
LI H, WANG R Y, GENG Z X, et al. Low illumination image enhancement algorithm based on multi-scale gradient domain guided filtering[J]. Computer Applications, 2019, 39(10): 3046-3052.
[8] RASHEED M T,et al. An empirical study on retinex methods for low-light image enhancement[J]. Remote Sensing, 2022, 14(18): 4608-4608.
[9] HU Yunxue, et al. Detail enhancement multi-exposure image fusion based on homomorphic filtering[J]. Electronics, 2022, 11(8): 1211-1211.
[10] JEON JONG JU, EOM IL KYU. Low-light image enhancement using inverted image normalized by atmospheric light[J]. Signal Processing, 2022: 108523.
[11] 崔圆斌, 田益民, 杜云飞, 等. 基于光照分量校正和补偿的低照度图像增强算法[J]. 数字印刷, 2021(6): 29-37.
CUI Y B, TIAN Y M, DU Y F, et al. Low illumination image enhancement algorithm based on illumination component correction and compensation[J]. Digital Printing, 2021(6): 29-37.
