计算机工程 ›› 2019, Vol. 45 ›› Issue (9): 253-259.doi: 10.19678/j.issn.1000-3428.0054281

• 图形图像处理 • 上一篇    下一篇

Contourlet域下基于多尺度特征的声呐图像分割

李鹏a,b, 陈嘉琦a,b, 马味敏a,b, 叶方跃a,b   

  1. 南京信息工程大学 a. 江苏省气象探测与信息处理重点实验室;b. 江苏省气象传感网络技术工程中心, 南京 210044
  • 收稿日期:2019-03-18 修回日期:2019-04-29 出版日期:2019-09-15 发布日期:2019-09-03
  • 作者简介:李鹏(1966-),男,教授、博士,主研方向为超声成像、图像处理;陈嘉琦、马味敏、叶方跃,硕士研究生。
  • 基金项目:
    国家自然科学基金(41075115);江苏省重点研发计划社会发展项目(BE201569);江苏省"六大人才高峰"第十一批高层次人才项目(2014-XXRJ-006)。

Sonar Image Segmentation Based on Multiscale Features in Contourlet Domain

LI Penga,b, CHEN Jiaqia,b, MA Weimina,b, YE Fangyuea,b   

  1. a. Jiangsu Key Laboratory of Meteorological Observation and Information Processing;b. Jiangsu Technology and Engineering Center of Meteorological Sensor Network, Nanjing University of Information Science and Technology, Nanjing 210044, China
  • Received:2019-03-18 Revised:2019-04-29 Online:2019-09-15 Published:2019-09-03

摘要: 水下环境复杂多变,导致声呐技术成像后的图像质量差,影响目标识别。为此,提出一种基于Contourlet域下多尺度高斯马尔可夫随机场(GMRF)模型的水平集声呐图像分割算法。采用Contourlet变换及逆变换获取声呐图像各尺度层下的纹理特征,通过GMRF对各层纹理特征建模,以描述局部结构空间信息并降低对噪声的敏感度。根据各层纹理特征模型,对声呐图像进行由粗到细尺度的水平集分割以得到分割结果。实验结果表明,该算法在不同声呐图像中的分割准确度超过90%,优于Otsu算法,且具有较低的复杂度和较强的鲁棒性。

关键词: Contourlet变换, 高斯马尔科夫随机场模型, 水平集, 声呐图像分割, 纹理特征

Abstract: Complex and changeable underwater environment leads to the poor quality of sonar images,decreasing the accuracy of target recognition.Therefore,a level set sonar image segmentation algorithm based on multiscale Gaussian Markov Random Field(GMRF) model under Contourlet domain is proposed.Contourlet transform and inverse transform are used to obtain the texture feature under each scale layer of the sonar image.The texture feature of each layer is modeled by GMRF to describe the local structure spatial information and reduce the sensitivity to noise.Based on the texture feature models of each layer,coarse-to-fine segmentation for level sets is performed on sonar images to obtain segmentation results.Experimental results show that the accuracy of the algorithm exceeds 90% in different sonar images,which is better than Otsu algorithm and has lower complexity and stronger robustness.

Key words: Contourlet transform, Gauss Markov Random Field(GMRF)model, level set, sonar image segmentation, texture feature

中图分类号: