J. Eur. Opt. Society-Rapid Publ. 21, 13( 2025) 145
Fig
. 2. Simulation analysis of PCA-NLM algorithm for ideal potassium lamp( a) Spectral characteristic peaks of potassium lamp;( b) Spectral characteristic peaks of potassium lamp with noise interference;( c) Interferogram;( d) Noisy spectrogram;( e) Spectrogram processed by PCA method;( f) Spectrogram processed by PCA-NLM Method.
The experimental results demonstrate that the PCA- NLM algorithm can significantly suppress noise and successfully extract the spectral characteristic peaks located at 766.49 nm and 769.89 nm.
3 Experimental system and measurement data
In the experiments, a potassium lamp was used to simulate the potassium combustion signal in the tail flame of an aircraft. In the scene of placing the potassium lamp against the sky background, the hybrid data from the potassium lamp and the sky background were collected via remote measurements. The experimental setup is shown in Figures 3a and 3b. The data was collected using SHS( HEP-756-S, Anhui Institute of Optics and Fine Mechanics, Chinese Academy of Sciences). The actual spectrometer is shown in Figure 3c. The main parameters of the instrument are listed in Table 1. The potassium lamp used was a JH-B cathode lamp from Beijing Shuguangming Electronic Lighting Instrument Co., Ltd.; the actual lamp is shown in Figure 3d.
In the experiments, different intensities of the tail flame radiation were simulated by changing the current of the potassium lamp. Due to limitations, six groups of interference data with different current intensities were used in this study. Table 2 lists the experimental parameters.
4 Data processing and analysis
Subsequently, under darkroom conditions in the laboratory, the experimental setup was configured as shown in Figure 3. Interference pattern data from a potassium lamp light source was collected, with the results presented in Figure 4a. After Fourier transform processing, the spectrum of potassium was successfully obtained, as shown in Figure 4b. However, the spectral signal obtained at this point was significantly affected by environmental disturbance noise. To further enhance the accuracy of potassium spectrum extraction, PCA method was employed for in-depth processing of the potassium spectrum, with the results displayed in Figure 4c. It clearly demonstrates that PCA can effectively distinguish between the potassium spectrum and environmental disturbances, but a trace amount of background noise remains difficult to completely eliminate. Next, based on the NLM denoising algorithm described in equations( 7)–( 9), further fine processing was conducted on the spectral data in Figure 4c, ultimately yielding the potassium lamp spectrum shown in Figure 4d. Experimental results indicate that the NLM denoising algorithm can further reduce the trace noise remaining after PCA. Meanwhile, Figure 4c clearly shows significant characteristic peaks at wavelengths of 766.49 nm and 769.89 nm. In subsequent analysis of mixed signals from the sky background, these two characteristic peaks will serve as key