New Algorithm to Improve High-Quality Spectral Signal

author: Susan

2022-01-07

New Algorithm to Improve High-Quality Spectral Signal

Challenge
However, when directly using a miniature spectrometer to detect trace multiple metal ion information in a high-concentration background, due to the uneven energy distribution of the light source, the large number of trace metal ions to be detected, and the selective absorption of different metal ions to different spectral bands. For other reasons, the spectral characteristics of the optical signal obtained by different integration sampling time parameters are very different, especially in the light source energy attenuation and the characteristic absorption band of the ion. If the integration time set by the micro-spectrometer is too large, the CCD will saturate, and the test data will be distorted. If the integration time is too small, the energy value will be weak, and the integration time parameter setting will affect.
According to the characteristics of the spectral signal, a high-quality spectral signal reconstruction algorithm based on binary search is proposed.
Algorithm steps (UV-Visible spectral signal reconstruction)

Set the instrument integration sampling parameter range [L_min, R_max], search for the wavelength range [min_index, max_index], and the integration time step to 1, based on this, collect only high-concentration zinc ion solutions under different integration times (reference solution ) Spectral energy signal and recorded in the matrix DATA;
Given the target reconstruction accuracy p and the target energy value max_top of the spectrum signal reconstructed by the specified target at each wavelength point, the setting range is [max_topL, max_topR], the step length is 1, and the initial value max_top=max_topL is set;
Calculate the reconstructed spectral signal θ of the target, and use the binary search algorithm to search and query the data in DATA to find out the target integration time of each wavelength point and the spectral energy value under this integration time;
Increase max_top by 1 and repeat step (3) until max_top=max_topR to obtain the reconstructed spectral signal under all target energy values;
Calculate the target reconstruction accuracy p value of the reconstructed spectral signal under all target energy values, and select the spectral energy value with the highest reconstruction accuracy as the information amount of signal reconstruction;
Reconstruct the spectral energy signal of the solution to be tested according to the integration time of the signal reconstruction of the reference solution at each wavelength point obtained in step (5); and calculate the reconstructed spectral absorbance signal according to formula (1) ;

.........................................(1)

Among them, A is the absorbance of the solution to be tested; I_O2 is the spectral energy signal of the solution to be tested in the sample tank including other metal ion solutions except zinc ions; I_O1 is the solution to be tested in the sample tank except for There is no spectral energy signal of other metal ion solutions except zinc ion.
Test Results
In order to verify the effectiveness of the proposed algorithm, the ATP2000P micro-spectrometer was used to collect ultraviolet-visible spectrum data. First, configure the mixed solution of Zn²⁺, Cu²⁺, Co²⁺ and Ni²⁺ concentration according to a certain experimental method. The concentration of the ions of the mixed solution is shown in Table 1. Among them, the serial number 1 solution is the reference solution, and the others are the solutions to be tested. Then set the parameters of the spectral signal reconstruction algorithm: p=96%, Lmin=1, R_max=21, min_Length=40, max_index=1184, max_topL=[0.7×216], max_topR=[0.8×216]; Finally, the ultraviolet-visible spectrum energy signal data is collected, and the spectral signal with more obvious and high-quality spectral characteristics is reconstructed according to the steps of the spectral signal reconstruction algorithm. [] means round down.

First, use ATP2000P to collect the spectral energy signal of the reference solution with an integration time of 1~21 ms, as shown in Figure 1.

Fig. 1 Spectral energy signals of reference solutions With different integration time

Then, according to the steps of this algorithm, the spectral energy signal of the reference solution is reconstructed for the target energy value of each wavelength point in the target range interval. Figure 2 is the spectral reconstruction of target signal of the reference solution with the highest reconstruction accuracy. Figure 3 is the spectral energy reconstruction signal of the reference solution with different integration times reconstructed according to the target signal in Figure 2. Figure 4 is the spectral energy reconstruction signal converted into absorbance.

Fig. 2 Spectral reconstruction of target signal of reference solution with different integral time Fig. 3 Spectral energy reconstruction signal of reference solution Fig.4 Comparison of spectral absorbance signal reconstruction before and after reconstruction
It can be seen from Figure 4 that the spectral characteristics of the pre-reconstruction and post-reconstruction spectral absorbance signals in the dashed frame are quite different. The spectral peak amplitude of the reconstructed spectral signal is large, and the spectral curve sensitivity is higher. It is easy to identify the ion species and ion concentration information in the wavelength range. For the ultraviolet-visible spectrum signal in the wavelength range of 280 to 760nm, the signal-to-noise ratio of the reconstructed absorbance signal is higher than the signal-to-noise ratio before reconstruction, which is conducive to the analysis and processing of low signal-to-noise ratio spectral data to a certain extent. It is also conducive to the analysis and extraction of the characteristics of the spectrum by the wavelength feature selection algorithm.

Fig. 5 Spectral energy reconstruction signals of all mixed solutions Fig. 6 Spectral absorbance reconstruction signals of all mixed solutions

Combine the signal data in Figure 5 and Figure 6 to calculate the performance indicators proposed in this article, and record the experimental results in Table 2. It can be seen from Table 2 that the signal-to-noise ratio of the reconstructed spectral signal is improved relative to the signal-to-noise ratio of the spectral signal before reconstruction, the reconstruction accuracy reaches an average of 94.84%, and the spectral features are also greatly enhanced.

Conclusion
Aiming at the problem of large differences in the characteristics of the obtained spectrum due to the selection of different integration time sampling parameters when the micro-spectrometer collects the spectrum signal of the trace polymetal ion concentration in the background of high concentration ratio, a high-quality spectrum based on binary search is proposed. Signal reconstruction algorithm. According to the spectral characteristics, two performance indexes of reconstruction accuracy and reconstruction feature saliency are defined, and then an ultraviolet-visible signal reconstruction method based on binary search is proposed. Taking the measured ultraviolet-visible spectrum signal of the mixed solution as an example, the experimental verification of signal reconstruction was carried out. Experimental results show that the average signal reconstruction accuracy can reach 94.48%, and the significance of reconstruction features is also improved. Under the condition that the noise of the reconstructed spectral signal is basically unchanged, the absorbance signal is enhanced, so the signal-to-noise ratio will also be improved by a small margin. From the perspective of signal processing, this method provides spectral signals with high signal-to-noise ratio and more obvious spectral characteristics for accurate spectral detection and analysis.
Related Products

ATP2000P Deuterium Halogen Light Source ATG1020H Cuvette Holders ATP0080 UV-Visible Fibers
Related Articles

“Spectroscopy and Spectral Analysis”
“High-Quality UV-Vis Spectrum Signal Reconstruction Algorithms Based on Binary Search”
“Science Technology and Engineering”