【摘要】 结合放射性能谱的统计特性,提出了一种基于高斯尺度空间的平滑滤波方法.构造由高斯函数构成的尺度空间,将能谱在该空间内进行多次投影,并在投影中调整高斯函数的权重以实现能谱平滑滤波.选取的标准方差σ和尺度j较小时(如σ=0.2～0.6,j=0),每次平滑后峰形畸变小,适于对弱峰的平滑;选取σ和j较大时(如σ=0.6～1.0,j=1,2,3),可使统计涨落得到较大抑制,并使谱峰成形加快,适于统计涨落较大且谱峰较明显的能谱平滑.结果表明,该方法可灵活地选择标准方差σ和尺度空间Vj,以适应平滑精度和处理速度的要求,且计算简便,是一种性能良好的平滑滤波方法.更多还原

【Abstract】 A smoothing filtering method based on the Gaussian scale-space was adopted according to the statistical properties of radioactive spectrum.The scale-space were constructed by the Gaussian functions,and the radioactive spectrum was projected for several times in this space.During the projection,the weights of Gaussian function were adjusted to achieve the radioactive spectrum smoothing.When the standard deviation σ and the scale space j are less,such as σ=0.2-0.6 and j=0,the distortion of peak shape is also less after each processing,which is suitable for the smoothing of weak peak.When σ and j are relatively larger,such as σ=0.6-1.0 and j=1,2,3,the statistical fluctuation can be inhibited greatly,and the forming of spectral peak can be accelerated,which is suitable for the smoothing of spectrum with relatively obvious peak and large statistical fluctuation.The results show that the standard deviation σ and the scale space Vj can be flexibly chose by this method to suit the smoothing precision and speed.This smoothing filtering method is of good performance and simple calculation.更多还原