Abstract: [Background] In the gamma energy spectrum measurement system, estimation and deduction of the baseline has been an important part of pulse amplitude measurement processing. With the digitalization of gamma energy spectrometer, the baseline estimation and deduction method also need to make corresponding improvement. [Purpose] Some relatively complex algorithms which with digital sampling of original pulses only can be realized on a computer, but not suitable for the digital systems which based on field programmable gate array (FPGA). Some algorithms which could be executed on FPGA didn't fully consider the influence of the normal pulse signal, which have poor real-time performance and accuracy. To solve this problem, paper puts forward a kind of new algorithm which could estimate and deduct the pulse baseline. The most important is that it can implement on FPGA. [Methods] This method is called moving average filter after excluding abnormal signal (MAEA). That is based on the original pulse signal, which obtained by analog-to-digital (A/D) sampling, pulses are viewed as abnormal signals, superimposed on the direct-current (DC) baseline, which need to be deducted. Then DC baseline is obtained by the moving average filtering method. While the pulse DC baseline is deducted by FPGA, the complexity of the instrument for subsequent processing will be lower. [Results] After experimental verification, this method has some characteristics:simple algorithm, well accuracy and high sensitivity. It not only could inhibit the signal's influence on the baseline such as low frequency interference and temperature drift, but also could track the change of the baseline real time and accurately and suppress spectral drift to a certain extent. [Conclusion] This method not only can be used for digital gamma energy spectrum measurement system, also can be used applied to other types, such as X-ray digital spectrometer.

Key words: Pulse baseline estimate and deduct, FPGA, Moving average filter