传统3D噪声模型将噪声分解为8个部分, 使复杂的噪声现象变得容易理解, 然而这个模型没有准确的噪声分离方法和物理机理阐述。因此提出一种改进的探测器噪声分离和量化方法：噪声频域分析方法。这种方法将多帧噪声数据在空间和时间上分离为11个部分, 并通过仿真实验分析各部分噪声对成像画面影响的强弱顺序, 同时指出了每部分噪声的物理机理。噪声频域分析为红外探测器建立了一个完整的噪声模型, 对理解红外探测器噪声特性、产生机理, 及有针对性地进行噪声处理指明了方向。

Traditional 3D noise model divides the noise into 8 parts, enabling the noise being easily understood. However, this model does not provide an accurate method to divide the noise parts and explanation of the noise generating mechanism. An optimized way to divide and quantize the detector noise was proposed by frequency analysis of noise. This noise analysis method divides noise into 11 parts by means of FFT, and then calculates the percent of each part in overall noise and analyzes the effect of each part on the image quality, and lastly explains the noise generating mechanism of each part of noise. The frequency analysis method provides a complete noise model for IR detectors and is proven to be an effective IR signal processing way.