摘要
气象卫星数据在采集或者传输的过程中,往往会引入不同程度的脉冲噪声。为提高图像质量,滤除噪音,进行可信度分析,提出了一种三阶段气象卫星数据脉冲噪声检测和滤除方法。在算法第一阶段,利用均值与均方差比值差值序列建立排序检测器(ROD),检测出图像中所有可能的脉冲噪声点;第二阶段,采用逐点滑动N×N窗口,按照每个可能的脉冲噪声点在不同的滑动窗口被重复检测到的次数,建立噪声可信度。第三阶段,对高噪声点进行滤波;对低噪声点迭代排序检测,如果可识别有噪声点,再进行滤波。结果表明,算法可以准确检测和滤除脉冲噪声,并保持数据中非噪声点信息。
Meteorological satellite data are often corrupted by pulse noises in the process of collection and trans-ission. A new approach has been proposed which involves noise detection and filtering based on a reliability analysis. The three - stage process begins with a rank - ordered detector (ROD) which takes the pulse noises based on the difference of sequence of mean and standard deviation ratio. This produces noise reliability statistics by moving an N x N window and taking values for each pixel N2 times, enabling us to identify possible noise values. A threshold is then established whereby the noise values can be sorted into high - reliability noise and low - reliability noise. Where high - reliability noise is identified, the value is filtered through a filter window during noise removal. Low reliability noise values are iterated and ranked so that upper - end values can be confirmed as true noise and thus discarded. The experiment shows that the filter can effectively remove impulse noises while retaining valid information with little or no distortion.
出处
《计算机仿真》
CSCD
北大核心
2009年第9期194-198,共5页
Computer Simulation
关键词
脉冲噪声
噪声检测
气象卫星数据
Pulse noise
Noise detection
Meteorological satellite data