摘要
针对云层中目标的识别问题 ,选择Sobel算子对输入图像进行了边缘检测 ,并以计算机模拟光学相关运算给出了识别结果。结果显示 ,在积状云和层状云背景中的目标识别较容易 ,在波状云中的目标识别较难。背景强度较强 ,目标亮度过强或过弱 ,或采集器件等因素导致图像中输入图像模糊 ,也会影响目标的识别率。另外 ,强目标相关峰比较尖锐 。
How to recognize targets under complicated background of cloud is studied in this paper by stimulating optical correlation. Some factors influencing recognition are calculated, such as the change of shape of cloud, target intensity, clearness or faintness of input image and so on. Simulation results show that targets can be found more easily in cumulus or stratus than in undulance cloud, because undulance cloud is more abrupt change in its intensity. The analyses also show that the clearer the input image is, the more easily the target is recognized;and the more intensive the target is, the more easily it is recognized.
出处
《激光技术》
CAS
CSCD
北大核心
2004年第2期133-136,140,共5页
Laser Technology
关键词
目标识别
光学相关
边缘检测
SOBEL算子
target recognition
optical correlation
edge detection
Sobel operator