-
摘要:
针对水下视觉同步定位与地图构建(SLAM)前端特征提取与特征匹配效果差的问题，提出一种用于水下视觉SLAM前端的基于加权融合的图像增强算法。该算法建立在2个图像的融合处理基础上，第1个图像经过基于自适应伽马校正和动态范围拉伸的水下图像亮度增强处理，第2个图像经过基于颜色判断和颜色补偿的灰度世界白平衡处理；通过计算2个图像的显著性权重和饱和性权重，对输入图像进行线性加权融合，得到最终的增强图像。通过水下彩色图像质量评价(UCIQE)和水下图像质量测量(UIQM)方法评估无参考水下增强图像质量，并用南卡罗来纳大学开源数据集测试所提算法的应用效果，结果表明：处理后，水下图像质量高、提取特征点数目多，可以显著提高水下视觉SLAM前端特征提取与特征匹配的效果。
Abstract:Aiming at the problem of poor feature extraction and feature matching in the underwater visual simultaneous localization and mapping (SLAM) front end, an image-enhanced algorithm based on weighted fusion is proposed for the underwater visual SLAM front end. Specifically, the algorithm is based on the fusion of two images: the second image is a white balance of gray world based on color judgment and color compensation, and the first image is an underwater image from brightness enhancement based on adaptive gamma correction and dynamic range expansion. Furthermore, the saliency weight and saturation weight of the two images are calculated, and the input images are linearly weighted and fused to obtain the final enhanced image.The system is tested using an open-source dataset from the University of South Carolina, and the improved underwater image quality is assessed using the techniques of underwater color image quality evaluation (UCIQE) and underwater image quality measurement (UIQM). Consequently, the results show that the processed image has the characteristics of high quality and a large number of extracted feature points, which can significantly improve the effect of front end feature extraction and feature matching of underwater visual SLAM.
-
图 11 直方图均衡化处理后水下图像SIFT特征匹配
Figure 11. Feature matching by SIFT of underwater images after histogram equalization
图 12 本文算法处理后水下图像SIFT特征匹配
Figure 12. Feature matching by SIFT of underwater images after the proposed algorithm
表 1 UCIQE无参考水下图像质量评估结果
Table 1. Assessment results of no-reference underwater image quality with UCIQE
图像来源 水下图像1 水下图像2 水下图像3 水下图像4 原始水下图像 3.3788 3.2122 5.4047 4.8687 直方图均衡化处理 4.0025 4.2486 5.0472 5.1449 本文算法 5.3088 5.0171 6.5764 6.1352 
下载: 导出CSV
表 2 UIQM无参考水下图像质量评估结果
Table 2. Assessment results of no-referenced underwater image quality with UIQM
图像来源 水下图像1 水下图像2 水下图像3 水下图像4 原始水下图像 0.0498 0.0698 0.2338 0.2118 直方图均衡化处理 0.4852 0.4524 0.6566 0.5556 本文算法 0.9575 1.0694 1.0917 1.0105
下载: 导出CSV
表 3 水下图像特征提取数
Table 3. Number of extracted feature points for underwater images
图像来源 水下图像1 水下图像2 水下图像3 水下图像4 原始水下图像 20 93 263 326 直方图均衡化处理 306 467 729 554 本文算法 620 780 4077 3560
下载: 导出CSV
表 4 水下图像正确的特征匹配对数
Table 4. Number of correct feature matching pairs for underwater images
图像来源 水下图像1、2 水下图像3、4 原始水下图像 7 15 直方图均衡化处理 39 17 本文算法 71 60
下载: 导出CSV
-
[1] 赵恒飞. 基于改进回环检测的VINS-Mono水下SLAM算法研究[D]. 杭州: 浙江大学, 2021.ZHAO H F. Research on VINS-Mono underwater SLAM algorithm based on improved loopback detection[D]. Hangzhou: Zhejiang University, 2021(in Chinese). [2] 王龑. 基于图像成像模型的水下图像增强方法研究[D]. 上海: 上海海洋大学, 2019.WANG Y. Research on underwater image enhancement method based on image imaging model[D]. Shanghai: Shanghai Ocean University, 2019(in Chinese). [3] 郭继昌, 李重仪, 张艳, 等. 面向水下图像的质量评价方法[J]. 中国图象图形学报, 2017, 22(1): 1-8.GUO J C, LI C Y, ZHANG Y, et al. Quality evaluation method for underwater images[J]. Chinese Journal of Image and Graphics, 2017, 22(1): 1-8(in Chinese). [4] ABDULLAH W M, KABIR M H, DEWAN M A A, et al. A dynamic histogram equalization for image contrast enhancement[J]. IEEE Transactions on Consumer Electronics, 2007, 53(2): 593-600. doi: 10.1109/TCE.2007.381734 [5] 魏郭依哲, 陈思遥, 刘玉涛, 等. 水下图像增强和修复算法综述[J]. 计算机应用研究, 2021, 38(9): 2561-2569.WEI G Y Z, CHEN S Y, LIU Y T, et al. Overview of underwater image enhancement and restoration algorithms[J]. Computer App- lication Research, 2021, 38(9): 2561-2569(in Chinese). [6] ZHANG J, ILA V, KNEIP L. Robust visual odometry in underwater environment[C]//Proceedings of the OCEANS-MTS/IEEE Kobe Techno-Oceans. Piscataway: IEEE Press, 2018: 1-9. [7] LAND E H. An alternative technique for the computation of the designator in the retinex theory of color vision[J]. Proceedings of the National Academy of Sciences, 1986, 83(10): 3078-3080. doi: 10.1073/pnas.83.10.3078 [8] 刘定通. 复杂背景下视频运动目标检测与跟踪算法研究[D]. 成都: 电子科技大学, 2016.LIU D T. Research on video moving target detection and tracking algorithm under complex background[D]. Chengdu: University of Electronic Science and Technology of China, 2016(in Chinese). [9] 汪秦峰. 基于直方图均衡化和Retinex的图像去雾算法研究[D]. 西安: 西北大学, 2016.WANG Q F. Research on image dehazing algorithm based on histogram equalization and Retinex[D]. Xi’an: Northwest University, 2016(in Chinese). [10] AL-HASHIM M, AL-AMEEN Z. Retinex-based multiphase algorithm for low-light image enhancement[J]. Traitement du Signal, 2020, 37(5): 733-743. doi: 10.18280/ts.370505 [11] 陈凌波, 朱树先. 基于结构和纹理感知的变分Retinex去雾算法[J]. 光学技术, 2021, 47(6): 733-740.CHEN L B, ZHU S X. Variational Retinex dehazing algorithm based on structure and texture perception[J]. Optical Technology, 2021, 47(6): 733-740(in Chinese). [12] ASHIBA M I, TOLBA M S, ELFISHAWY A S, et al. Gamma correction enhancement of infrared night vision images using histogram processing[J]. Multimedia Tools and Applications, 2019, 78(19): 27771-27783. doi: 10.1007/s11042-018-7086-y [13] CHANG Y, JUNG C, KE P. Automatic contrast-limited adaptive histogram equalization with dual gamma correction[J]. IEEE Access, 2018, 6: 11782-11792. doi: 10.1109/ACCESS.2018.2797872 [14] 黄国祥. RGB颜色空间及其应用研究[D]. 长沙: 中南大学, 2002.HUANG G X. RGB color space and its application research[D]. Changsha: Central South University, 2002(in Chinese). [15] MAX K. Computer graphics and geometric modeling: Implementation and algorithms[M]. Berlin: Springer, 2005. [16] 马玲, 张晓辉. HSV颜色空间的饱和度与明度关系模型[J]. 计算机辅助设计与图形学学报, 2014, 26(8): 1272-1278.MA L, ZHANG X H. The relationship model of saturation and lightness in HSV color space[J]. Journal of Computer Aided Design and Graphics, 2014, 26(8): 1272-1278(in Chinese). [17] 许冰, 牛燕雄. 复杂动态场景下目标检测与分割算法[J]. 北京航空航天大学学报, 2016, 42(2): 310-317.XU B, NIU Y X. Target detection and segmentation algorithm in complex dynamic scenes[J]. Journal of Beijing University of Aeronautics and Astronautic, 2016, 42(2): 310-317(in Chinese). [18] ACHANTA R, HEMAMI S, ESTRADA F, et al. Frequency-tuned salient region detection[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE Press, 2009: 1597-1604. [19] LYU W, LU W, MA M. No-reference quality metric for contrast distorted image based on gradient domain and HSV space[J]. Journal of Visual Communication and Image Representation, 2020, 69: 102797. doi: 10.1016/j.jvcir.2020.102797 [20] 何可丁, 王庆. 一种自适应的白平衡解决方案研究[J]. 齐鲁工业大学学报, 2021, 35(2): 1-7.HE K D, WANG Q. Research on an adaptive white balance solu- tion[J]. Journal of Qilu University of Technology, 2021, 35(2): 1-7(in Chinese). [21] YANG M, SOWMYA A. An underwater color image quality evaluation metric[J]. IEEE Transactions on Image Processing, 2015, 24(12): 6062-6071. doi: 10.1109/TIP.2015.2491020 [22] PANETTA K, GAO C, AGAIAN S. Human-visual-system-inspired underwater image quality measures[J]. IEEE Journal of Oceanic Engineering, 2015, 41(3): 541-551. -
下载:
点击查看大图
计量
- 文章访问数: 553
- HTML全文浏览量: 136
- PDF下载量: 59
- 被引次数: 0
