摘要
在高频声纳图像目标检测中,对图像分割后,需要对前景目标参数进行提取。水下声学图像相对于光学图像而言,分辨率相对较低,并且通常不会有特别复杂的图形图案进行处理。本文根据声学图像的特点,提出了一种连通成分标记算法,利用此算法可以对分割后的声纳目标进行标记,并快速提取出声图中的目标个数,以及各个目标的位置、面积等特征参数。此算法在VC++软件平台上对扫描声纳图像进行了实时处理,结果验证了该算法的可行性。
In object detection of high frequency sonar image, it needs to get the parameters of foreground objects after image segmentation. Compared with optical image, the sonar image resolution is lower, and the sonar image generally has no very complicated graphics to process. This paper describes a connected components labeling algorithm, which can quickly gets the number of objects in the processing sonar image, and each object’s position, area and other features. This algorithm has been realized on VC ＋＋ , and used in sonar image processing in real time.
出处
《微计算机信息》
2010年第2期16-17,共2页
Control & Automation
关键词
连通成分
标记算法
声纳图像
目标检测
Connected Component
Labeling Algorithm
Sonar Images
Object Detection
