During the data acquisition of an extreme ultraviolet (EUV) spectroscopic diagnostic system in experimental advanced superconducting Tokamak (EAST), significant amounts of one-pixel noises are consistently observed. This is attributed to the influence of hard X-ray on the charge coupling device (CCD) detector.

This study aims to detect and denoise spectral images in the EUV spectral image processing system based on the field programmable gate array (FPGA).

First of all, based on limiting filtering algorithm, spectral image processing was optimized by replacing fixed limiting thresholds and sample deviations using parameters such as standard deviation and deviation from the mean, and Andor Solis software to was applied to converting the SIF format spectra to BMP format spectral images. Then, data discrimination method was combined with limiting filtering algorithm to process spectral image data in stages according to the setting working area of the algorithm. Finally, a simulation test module was designed to process the video data converted from EUV spectral images in AX7Z100 ZYNQ FPGA platform, simulating the actual acquisition process for functional testing of the system. To assess the system's capacity to protect effective spectral data, the image data was transferred to one-dimensional spectral data for comparative analysis.

Based on the experimental results, the EUV spectral image processing system of this study can effectively eliminate noise data points in the spectral image while essentially maintaining the integrity of the spectral data.

This study enhances the processing of the spectral data and provides a new technological path for EUV spectral image processing.