摘要
针对典型海空目标的识别要求,介绍了一种基于统计学的目标识别方法。该方法通过对图片、视频或实时采集的图像进行预处理、目标分割、边界搜索、内部填充等处理后,提取目标的统计学特征,根据目标的特征值和数据库中的已有样本对未知目标进行识别。对5种舰船三维模型在不同方位角和俯仰角的平面投影图像进行了实验,识别概率能达到80%以上。实验结果表明,在被测目标样本充足的情况下,该方法能达到较高的识别率。
A method of target identification based on statistics is presented according to the identification demand to the typical sea and sky targets. After obtaining an image, the process of filtering, image segmentation, edge searching, interior filling etc., are done, and the target statistic features are extracted. According to the feature values and the sample database, the unknown target can he identified. The experiment on five ship-models indicates that the identification probability can reach up to 80 percent.
出处
《光学与光电技术》
2009年第3期67-70,共4页
Optics & Optoelectronic Technology
关键词
统计学
目标识别
特征值
样本
statistics
target identification
feature value
sample
