摘要
提出了一种大尺度卫星遥感图像中基于知识的快速识别港口方法。首先,根据港口目标的知识,详细分析了其分布特征和固有特征;然后,针对港口大小的不同分别设计两类港口候选区域提取方法;再采用带有错误控制策略的阈值方法进行海洋和陆地的分割;最后,利用港口的固有特征(半封闭区域),实现港口识别。
A recognition method of harbor target in large gray remote sense image was presented. Distributed inherit characters were analyzed according to the knowledge of harbor targets. Two methods of extracting the candidate region of harbor were designed in accordance with different sizes of the harbors. Thresholds were used to segment the land and the sea with strategies of controlling the segmenting error. Finally, harbor recognition was implemented according to its inherent character ( half close region of seawater).
出处
《计算机应用》
CSCD
北大核心
2006年第3期729-732,共4页
journal of Computer Applications
关键词
多尺度
候选区域
特征提取
门限分割
multi-scale
candidate region
character extraction
threshold segment