第 23 卷 第 1 期
2017年2月
(自然科学版)
JOURNAL OF SHANGHAI UNIVERSITY (NATURAL SCIENCE)
Vol. 23 No. 1
Feb. 2017
DOI: 10.3969/j.issn.1007-2861.2016.07.018
基于结构森林边缘检测和 Hough 变换的
海天线检测
徐良玉
1
, 马录坤
1
, 谢 燮
1
, 彭 艳
1
, 彭艳青
2
, 崔建祥
1
(1. 上海大学 机电工程与自动化学院, 上海 200072; 2. 中国人民解放军理工大学 理学院, 南京 210007)
摘摘摘要要要: 海天线是海面环境图像所具有的重要特征之一, 海天线的检测对划分海空、海界区域
以及目标检测有重要作用. 提出了一种结合结构森林快速边缘检测和概率 Hough 变换的海天
线检测方法. 首先通过高斯低通滤波来减小海面浪纹、光照反射等局部纹理影响, 然后使用已
完成训练的结构化随机森林为 每个像素贴上边缘标签——二 值化, 最后通过 Hough 变换原理
拟合海天线. 实验结果表明, 该方法可以较好地忽略局部干扰边缘, 强化边界提取, 对复杂海天
背景下的海天线检测具备鲁棒性和高准确性.
关关关键键键词词词 : 海天线检测; 结构化随机森林; 决策树; 边缘检测; Hough 变换
中中中图图图分分分类类类号号号 : TP 242.62 文文文献献献标标标志志志码码码 : A 文文文章章章编编编号号号 : 1007-2861(2017)01-0047-09
Sea-sky line detection based on structured forests edge
detection and Hough transform
XU Liangyu
1
, MA Lukun
1
, XIE Xie
1
, PENG Yan
1
,
PENG Yanqing
2
, CUI Jianxiang
1
(1. School of Mechatronic Engineering and Automation, Shanghai University, Shanghai 200072, China;
2. College of Sciences, PLA University of Science and Technology, Nanjing 210007, China)
Abstract: The sea-sky line is an important feature in the sea-surface environment image,
and detection of the sea-sky line is essential in dividing the sea and sky, and detecting
the coastline area and objects. This paper provides a method to detect the sea-sky line
using structured forests edge detection and Hough transform. The method uses a Gaussian
low-pass filter to reduce the influence of regional textures such as wave texture and light
reflection. A trained structured random decision forest is then used to label each pixel,
and binarize it to determine whether it belongs to an edge or not. Hough transform is used
to fit the sea-sky line more accurately. Experimental results show that this method can
neglect clutter edge, greatly improve edge detection, and effectively extract sea-sky lines
from a complicated sea-sky background with high robustness and accuracy.
Key words: sea-sky line detection; structured random forest; decision tree; edge detection;
Hough transform
收稿日期: 2016-12-29
基金项目: 国家自然科学基金资助项 目(61 403 245) ; 上海市自然科学基金资助项目(13ZR1454300); 上海市科委能力
建设资助项目(14500500400)
通信作者: 彭 艳(1982—), 女, 副教授, 研究方向为无人艇导航和控制及其总体技术. E-mail: pengyan@shu.edu.cn