摘要
研究了为水下机器人声视觉设计的一套三维成像声纳系统的目标识别方法,该声纳可以在一次发射中获取一系列二维切片图像。根据面阵水平、垂直分辨率的不同和图像中目标的切片数量,结合水声环境对水下成像的影响,分析了几种不同目标切片图像的特点,提出一种以直方图中目标能量为特征的识别方法,该方法在算法实现上避免了三维重建的步骤,使识别速度显著提高,在对水池数据和海试数据的分类识别试验中证明其能有效地识别水下目标。
This paper discusses the target recognition with a sonar 3D imaging system for AUV. A set of progressive 2D sliced images from one acoustic power transmission is first obtained from the system. Then according to the difference of the horizontal and vertical resolution and the slices＇ number, a histogram method is proposed for the recognition to avoid a 3-D reconstruction. The classification experiments from the pool and ocean show that the presented system can perform well.
出处
《自动化技术与应用》
2008年第2期55-57,共3页
Techniques of Automation and Applications
关键词
切片图像
直方图
图像分割
目标识别
sliced image
energy histogram
image segmentation
target recognition
