侧扫声呐图像特征提取改进方法

王其林; 王宏健; 李 庆; 肖 瑶; 班喜程

doi:10.11993/j.issn.1673-1948.2019.03.010

侧扫声呐图像特征提取改进方法

doi: 10.11993/j.issn.1673-1948.2019.03.010

哈尔滨工程大学自动化学院, 黑龙江哈尔滨, 150001

基金项目: 国家自然科学基金重点项目(61633008); 国家自然科学基金(青年项目)(51609046); 黑龙江省自然科学基金(F2015035)

详细信息

作者简介:
王其林(1992-), 男, 在读硕士, 主要研究方向为船舶航行与作业控制理论与工程.

中图分类号: TJ6; TP751
计量
- 文章访问数: 436
- HTML全文浏览量: 0
- PDF下载量: 243
- 被引次数: 0
出版历程
- 收稿日期: 2018-11-04
- 修回日期: 2018-12-18
- 刊出日期: 2019-06-30

Improved Feature Extraction Method for Side Scan Sonar Images

College of Automation, Harbin Engineering University, Harbin 150001, China

摘要

摘要: 由于侧扫声呐图像对于研究海底状况有着重要的作用, 因此常采用点特征和线特征提取来分析声呐图像的特点。其中点特征提取分析采用尺度不变特征变换(SIFT)算法, 直线特征提取采用Hough变换和线段检测(LSD)算法。文中针对Hough变换在提取直线特征时因基本空间转换方式存在不足而使得提取的线特征不理想的问题, 提出了一种效果更好的空间转换方式, 该方法需要建立2个图像边缘掩码矩阵, 一个保持原状用于寻找直线特征, 另一个根据边缘点转化情况相应减少对应边缘点, 这种方法的设计可以保证属于直线的特征点都能参与到直线特征的形成中; 针对LSD算法中检测到的直线特征会因相交而出现断裂的问题, 设计了一种断裂线特征拟合方法, 利用角度、位置等条件将被判定为断裂的直线特征进行拟合, 减少因直线特征断裂造成直线过短过多的问题, 然后通过长度阈值筛选出合适的直线特征。通过优化前后的方法对侧扫声呐图像进行直线特征提取验证, 证明了所改进的Hough变换可以更好地提取到线特征, 而优化后的LSD算法也减小了线特征断裂的影响。同时对于待处理的声呐图像, 点特征提取数量较多, 不能很好地体现图像内容特点, 而改进之后的线特征可以提取出图像内容的轮廓, 信息丰富, 能更好地表现了图像的内容。以上改进方法可为有关线特征提取研究提供参考。
- 侧扫声呐 /
- 特征提取 /
- 尺度不变特征变换算法 /
- Hough变换 /
- 线段检测算法
Abstract: To solve the problem in feature extraction of side scan sonar images that the line feature extracted by Hough transform is not satisfactory due to the deficiency of the basic space conversion method, a better space conversion method is proposed. This method needs to establish two image edge mask matrices — one for maintaining the original shape to find the straight line features, and the other for reducing the corresponding edge points according to the edge point transformation. This method can ensure that the feature points belonging to the straight line participate in the formation of linear features. However, the linear features detected in the line segment detection(LSD) algorithm will be broken due to intersection, hence a method of fitting the broken line features is designed, in which the straight line features being determined as ‘broken’ are simulated via angle and position so as to deal with the trouble of having too short or too much straight lines due to the fracture of the linear features, and then the appropriate straight line features are selected according to the length threshold. The linear feature extraction of side scan sonar images verifies that the improved Hough transform can extract the line features better, and the optimized LSD algorithm can reduce the influence of line feature fracture. For the sonar image to be processed, the number of point feature extraction is large, which can not reflect the characteristics of the image content well; and the improved line feature can extract the outline of the im-age content, and the information is rich, which better expresses the content of the image. It is provide reference for relevant line feature extraction research.
- side scan sonar /
- feature extraction /
- scale invariant feature transform(SIFT) algorithm /
- Hough transform /
- line segment detection(LSD) algorithm

HTML全文

参考文献(23)

[1]	Wei S, Leung H, Myers V. An Automated Change Detection Approach for Mine Recognition Using Sidescan So-nar Data[C]//Proceedings of the 2009 IEEE International Conference on Systems, Man, and Cybernetics. San Antonio, TX, USA: IEEE, 2009: 553-558.
[2]	Reed S, Petillot Y, Bell J. An Automatic Approach to the Detection and Extraction of Mine Features in Sidescan sonar[J]. IEEE Journal of Oceanic Engineering, 2003, 28(1): 90-105.
[3]	Rao C, Mukherjee K, Gupta S, et al. Underwater Mine Detection Using Symbolic Pattern Analysis of Sidescan sonar images[C]//2009 American Control Conference. St. Louis, MO, USA: IEEE, 2009: 5416-5421.
[4]	罗进华, 蒋锦鹏, 朱培民. 基于数学形态学的侧扫声呐图像轮廓自动提取[J]. 海洋学报, 2016, 38(5): 150-157. Luo Jin-hua, Jiang Jin-peng, Zhu Pei-min. Automatic Ex-tration of the Side-scan Sonar Imagery Outlines Based on Mathematical Morphology[J]. Acta Oceanologica Sinica, 2016, 38(5): 150-157.
[5]	李海森, 高珏, 徐超. 应用TBD的多波束声呐图像序列SIFT特征追踪[J]. 哈尔滨工程大学学报, 2016, 37(8): 1009- 1014. Li Hai-sen, Gao Yue, Xu Chao. SIFT Feature Tracking in Multi-beam Sonar Image Sequences Using the Trackbefore-detect Technique[J]. Journal of Harbin Engineering University, 2016, 37(8): 1009-1014.
[6]	王力. 基于直线特征的异源景象匹配技术研究[D]. 北京: 中国科学院大学, 2015.
[7]	黄勇. 基于相关点与几何约束的直线匹配方法研究[D].南昌: 南昌航空大学, 2016.
[8]	Alcantarilla P F, Bergasa L M, Davison A J. Gauge-SURF Descriptors[J]. Image and Vision Computing, 2013, 31(1): 103-116.
[9]	Von K T. Sedaghat A, Ebadi H. Remote Sensing Image Matching Based on Adaptive Binning SIFT Descriptor[J]. IEEE Transactions on Geoscience and Remote Sensing, 2015, 53: 5283-5293.
[10]	Zhao J, Guo H, Wu J. Attribute Reduction for SIFT Local Descriptors Using PCA and CAIM[C]//International Congress on Image and Signal Processing, Liaoning, Anshan: IEEE, 2014.
[11]	傅卫平, 秦川, 刘佳. 基于SIFT算法的图像目标匹配与定位[J]. 仪器仪表学报, 2011, 32(1): 163-166. Fu Wei-ping, Qin Chuan, Liu Jia. Matching and Location of Image Object Based on SIFT Algorithm[J]. Chinese Journal of Scientific Instrument, 2011, 32(1): 163-166.
[12]	Yahyanejad S, Bernhard R. A Fast and Mobilesystem for Registration of Low-altitude Visual and Thermal Aerial Images Using Multiple Small-scale UAVs[J]. ISPRS Journal of Photogrammetry and Remote Sensing, 2015, 104(9): 189-202.
[13]	Amintoosi M, Fathy M, Mozayani N. A Fast Image Registration Approach Based on SIFT Key-points Applied to Super-resolution[J]. The Imaging Science Journal, 2012, 60(4): 185-201.
[14]	Marr D, Hildreth E. Theory of Edge Detection[C]//Proceedings of the Royal Society of London B, Biological Sciences, Britain: IEEE, 1980.
[15]	Olga B, Victor L, Pushmeet K. On Detection of Multiple Object Instances Using Hough Transforms[J]. IEEE Tra- nsactions on Pattern Analysis and Machine Intelligence, 2012, 34(9): 73-84.
[16]	滕今朝, 邱杰. 利用Hough变换实现直线的快速精确检测[J]. 中国图象图形学报, 2008(2): 234-237. Teng Jin-zhao, Qiu Jie. Fast and Precise Detection of Straight Line with Hough Transform[J]. Journal of Image and Graphics, 2008(2): 234-237.
[17]	刘桂雄, 申柏华, 冯云庆. 基于改进的Hough变换图像分割方法[J]. 光学精密工程, 2002(3): 257-260. Liu Gui-xiong, Shen Bo-hua, Feng Yun-qing. Study of Image Segmentation Based on Improved Hough Transform[J]. Optics and Precision Engineering, 2002(3): 257-260.
[18]	李曙光. 声呐图像的特征提取技术研究[D]. 哈尔滨: 哈尔滨工程大学, 2014.
[19]	李庆斌, 石丹, 霍冠英. 基于Contourlet变换的海底声呐图像特征提取与分类[J]. 海洋学报, 2011, 33(5): 163-168. Li Qing-bin, Shi Dan, Huo Guan-ying. Feature Extraction and Classification of Seabed Sonar Images Based on Con- tourlet transform[J]. Acta Oceanologica Sinica, 2011, 33 (5): 163-168.
[20]	韩秋蕾, 朱明, 姚志军. 基于改进Hough变换的图像线段特征提取[J]. 仪器仪表学报, 2004, 25(4): 436-439. Han Qiu-lei, Zhu Ming, Yao Zhi-jun. Line Segment Feature Extration Based on Improved Hough Transform[J]. Chinese Journal of Scientific Instrument, 2004, 25(4): 436-439.
[21]	王鑫. 水下环境中图像匹配算法的研究[D]. 太原: 太原理工大学, 2014.
[22]	路春平. 基于几何特征的直线匹配方法研究[D]. 南昌:南昌航空大学, 2014.
[23]	罗超, 熊国飞. 基于LSD的低空影像线特征提取[J]. 科技广场, 2016(11): 16-22. Luo Chao, Xiong Guo-fei. Low Altitude Image Line Feature Extraction Based on LSD[J]. Science Mosaic, 2016 (11): 16-22.