随着科技的发展，适用于结构分析的传统单能X射线计算机层析(CT)成像技术，已不能满足现代工业对物质组分区分与鉴定的功能成像需求。这是由于在X射线CT系统中，现有重建算法的单能假设与CT投影的多谱性不一致，导致CT重建质量差，无法组分区分。基于光子计数探测器的能谱分离成像思想，提出了基于能谱滤波分离的多谱CT成像方法，该方法通过在X射线发射端加滤波片的方式，实现能谱滤波分离，并通过变能量成像，获得近似单能的递变能量投影序列；针对滤波后噪声水平较高问题，利用EM-TV重建算法，实现了多谱CT成像，可满足组分区分的需求。仿真实验结果表明，对于密度相近的检测对象，该方法可以满足组分区分的要求。

With the development of technology, the traditional single-energy X-ray computed tomography (CT) imaging technology can′t meet the modern industry needs of functional imaging for substance distinction and identification because of the inconsistency between the projection acquisition process and the reconstruction assumption. A multi-spectrum CT imaging method based on energy spectrum filtering separation in view of photon counting imaging detector is presented. Spectrum filtering separation is realized by filtering at X-ray source and monochromatic projection sequence is gotten by various energy imaging. Then an EM-TV algorithm is used at the condition of low dose. By this method, mult-spectrum imaging is realized to meet the demand of substance distinction. Simulation result, which aims at components whose densities are similar, shows that this way can meet the requirement of component discrimination.