测绘通报 ›› 2021, Vol. 0 ›› Issue (11): 48-53.doi: 10.13474/j.cnki.11-2246.2021.337

• 学术研究 • 上一篇    下一篇

基于霍夫直线检测和区域生长的合成孔径声呐图像线目标提取

欧阳犬平1,2, 周立2, 王圣2, 宋士林3, 芦斌2, 冯俊俊2   

  1. 1. 武汉长江航道救助打捞局, 湖北 武汉 430014;
    2. 江苏海洋大学海洋技术与测绘学院, 江苏 连云港 222005;
    3. 国家海洋局北海海洋技术保障中心, 山东 青岛 266033
  • 收稿日期:2020-11-02 出版日期:2021-11-25 发布日期:2021-12-02
  • 通讯作者: 周立。E-mail:zhoulilyg@aliyun.com
  • 作者简介:欧阳犬平(1994-),男,硕士,主要研究方向为海洋空间信息技术。E-mail:2994261063@qq.com
  • 基金资助:
    国家重点研发计划“海洋环境安全保障”重点专项(2018YFC1405702);海岸带地理环境监测国家测绘地理信息局重点实验室开放基金(BE2017125);连云港国际科技合作计划(HZ201907)

Line target extraction of synthetic aperture sonar image based on Hough line detection and region growing

OUYANG Quanping1,2, ZHOU Li2, WANG Sheng2, SONG Shilin3, LU Bin2, FENG Junjun2   

  1. 1. Wuhan Changjiang Waterway Rescue and Salvage Bureau, Wuhan 430014, China;
    2. School of Marine Technology and Marine Technology and Geomatics, Jiangsu Ocean University, Lianyungang 222005, China;
    3. North China Sea Marine Technical Support Center, State Oceanic Administration, Qingdao 266033, China
  • Received:2020-11-02 Online:2021-11-25 Published:2021-12-02

摘要: 本文针对合成孔径声呐图像线目标提取过程中,由于线目标宽度(直径)小且图像受相干斑噪声影响严重而导致的提取困难的问题,提出了基于霍夫直线检测和区域生长的线目标提取算法。首先利用Otsu算法计算区域生长终止阈值,并对噪声进行初步抑制;然后在对图像进行霍夫直线粗检测的基础上,对检测结果进行不均匀分割,再进行细检测,得到区域生长种子点;最后利用区域生长对线目标进行还原。试验结果表明,该算法能准确地提取线目标;与传统区域生长对比,该算法选取的种子点数量提升了5.37倍,且线目标提取结果具有更好的连续性和完整性。

关键词: 合成孔径声呐, 线目标提取, 霍夫直线检测, 区域生长, Otsu算法

Abstract: A line target extraction algorithm based on Hough line detection and region growing is proposed to solve the problem of small width (diameter) of line target and serious influence of speckle noise on the image in the process of synthetic aperture sonar image line target extraction. Firstly, Otsu algorithm is used to calculate the growth termination threshold of the region and preliminarily suppress the noise; Secondly, based on the Hough line roughness detection of the image, the detection results are divided unevenly, and then the fine detection is carried out to obtain the seed points growing in the region; Finally, region growing is used to restore the line target. Experimental results show that the algorithm can accurately extract line targets, and the number of seed points selected by the algorithm is increased by 5.37 times compared with the traditional region growth, and the line target extraction results have better continuity and integrity.

Key words: synthetic aperture sonar, line target extraction, Hough line detection, region growing, Otsu algorithm

中图分类号: