摘要
针对白天测星的强噪声背景下弱小星点目标的特点,提出了基于USAN原则的检测方法,及邻域聚类法和星光成像原理相结合的虚警目标点的剔除方法。试验证明图像处理检测时间在40 ms以内,满足系统实时性要求。当SNR大于3．5时,成功提取概率大于99%。该算法提高了系统白天测星的能力。
A univalue seggment assimilating nucleus method for day and night star is proposed according to the low SNR. Based on neighborhood clustering and starlight imaging theory, an algorithm to remove noise is given. The experimental results prove that the image processing and detecting speed is less than 40 ms. If the SNR is upto 3.5, the detecting probability gets to 99%.
出处
《光学与光电技术》
2006年第3期56-58,共3页
Optics & Optoelectronic Technology
关键词
吸收同值核区
聚类
星体检测
门限分割
univalue segment assimilating nucleus
clustering
star detecting
threshold segmentation
