Miller T R, Sampathkumaran K S
Eur J Nucl Med. 1982;7(1):22-7. doi: 10.1007/BF00275240.
The finite impulse response (FIR) digital filter is a spatial domain filter with a frequency domain representation. The theory of the FIR filter is presented and techniques are described for designing FIR filters with known frequency response characteristics. Rational design principles are emphasized based on characterization of the imaging system using the modulation transfer function and physical properties of the imaged objects. Bandpass, Wiener, and low-pass filters were designed and applied to 201Tl myocardial images. The bandpass filter eliminates low-frequency image components that represent background activity and high-frequency components due to noise. The Wiener, or minimum mean square error filter 'sharpens' the image while also reducing noise. The Wiener filter illustrates the power of the FIR technique to design filters with any desired frequency response. The low-pass filter, while of relative limited use, is presented to compare it with a popular elementary 'smoothing' filter.
有限脉冲响应（FIR）数字滤波器是一种具有频域表示的空间域滤波器。本文介绍了FIR滤波器的理论，并描述了设计具有已知频率响应特性的FIR滤波器的技术。基于使用调制传递函数对成像系统的表征以及成像对象的物理特性，强调了合理的设计原则。设计了带通、维纳和低通滤波器，并将其应用于201Tl心肌图像。带通滤波器消除了代表背景活动的低频图像成分和由噪声引起的高频成分。维纳滤波器，即最小均方误差滤波器，在降低噪声的同时“锐化”图像。维纳滤波器展示了FIR技术设计具有任何所需频率响应的滤波器的能力。低通滤波器虽然用途相对有限，但也被提出来与一种流行的基本“平滑”滤波器进行比较。
