摘要
水下目标识别一直是目标识别领域的研究热点,为此,以水雷目标识别为需求背景提出了一种基于视觉的水雷目标识别方法。该方法从水雷目标形状的本质特征出发,根据几何矩的物理意义,结合区域特征和边界特征构造了三个适合于水雷目标形状描述子,采用阈值判决的方法,实现水雷目标的识别。实验结果表明,该方法比基于不变矩的方法识别率更高,具有较好的稳定性,尤其适用于水下特定形状的目标识别和目标受到部分遮挡的情况。其中水下图像处理和特征的定义方法对水下目标识别具有指导和借鉴意义。
Underwater object recognition has long been the research focus in the field of object recognition.For mine object recognition,we propose a novel method based on vision.Based on the essential features of the mines' shape,the method proposes three shape feature descriptors of the mines.According to the physical significance of the geometric moments,it combines the regional characteristics and boundary characteristics,and employs threshold decision to complete mine object recognition.The experimental results show that the recognition rate of the method is much higher than that of the methods based on invariant moments.The method has good stability,especially suitable for the specific shape of the underwater object recognition and partial occlusion of the object.In the method,underwater image processing and feature definition approaches provide guidance and reference for underwater object recognition.
出处
《海洋工程》
CSCD
北大核心
2012年第4期154-160,共7页
The Ocean Engineering
基金
中国科学院知识创新工程重大资助项目(YYYJ-0917)
关键词
水下机器人
水下图像
目标形状
不变矩
水雷识别
视觉
autonomous underwater vehicle
underwater image
object shape
invariant moments
mine recognition
vision