Open Access
Add to BibSonomy
Abstract
This study introduces an advanced approach for assessing the damage state of charge-coupled devices (CCDs) caused by laser interactions, leveraging a multi-source and multi-feature information fusion technique. We established an experimental system that simulates laser damage on CCDs and collects diverse data types including echo information from active laser detection based on the ‘cat's eye’ effect, plasma flash data, and surface image characteristics of the CCD. A probabilistic neural network (PNN) was utilized to integrate these data sources effectively. Our analysis demonstrated that using multiple features from single sources significantly improves the accuracy of the damage assessment compared to single-feature evaluations. The error rates using dual features from each information type were 10.65% for cat's eye echo, 7.3% for plasma flash, and 7.17% for surface image analysis. By combining all three information sources and six features, we successfully reduced the error rate to 0.85%, with the evaluation time under 60 milliseconds. These findings confirm that our multi-source, multi-feature fusion method is highly effective for the online and real-time evaluation of CCD damage, offering significant improvements in the operational reliability and safety of devices in high-energy environments.
© 2024 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
As the ‘eyes’ of reconnaissance equipment, photoelectric loads are often used for information acquisition, situation awareness and guidance [1–4]. Charge-coupled Device (CCD) is widely used in photoelectric load because of its high imaging quality, wide dynamic range and high resolution. As a photoelectric imaging detector, CCD is located in the focal plane of optical focusing system, and it is easy to be blinded by thermal damage caused by focused intense laser, so damaging CCD by intense laser becomes one of the main means of photoelectric countermeasures [5–7].
The damage mechanism caused by intense laser on CCD has been widely studied. Experiments such as CCD interference, point damage, line damage, surface damage and blindness caused by different wavelengths, different working mechanisms and different energies have been carried out by many researchers, and different damage thresholds have been measured [8–11]. CCD damaged by high power laser already can be used in the field of photoelectric countermeasures, but the problem of damage effect evaluation still needs to be solved.
In photoelectric countermeasure, whether the damage state of CCD can be evaluated correctly and effectively is an important factor to determine the efficiency of photoelectric countermeasure. At present, the commonly used methods that can be used for on-line evaluation of CCD damage state include scattered light intensity method, plasma flash method, scattered light polarization method, speckle interferometry method, image discrimination method and so on [12]. Among them, the method of detecting and evaluating the CCD surface morphology based on the ‘cat's eye’ echo intensity information is the most commonly used [13–17], and echo polarization information [18,19], plasma flash information [20] and image information [21,22] are also adopted. For example, Zhang et al. [16] established the lidar cross-section model of the optical system echo of the defocused ‘cat's eye effect’ target under the conditions of normal incidence and oblique incidence and provided the quantitative index for detecting the ‘cat's eye’ target. Lei et al. [17] established an experimental system of ‘cat's eye’ echo detection to study the damage evolution of silicon on the focal plane under laser irradiation by using the change of echo image. Qin et al. [18] established a polarized bidirectional reflection distribution function model, and studied the influence of the surface roughness of ‘cat's eye’ target on the polarization degree of echo scattering. Hu et al. [19] developed a polarization imaging detection system for obtaining the surface damage state of photoelectric detector, which proved that the polarization parameter of ‘cat's eye’ echo can effectively deduce the surface damage state of photoelectric detector in real time; Ge et al. [20] collected the spectrum of plasma flash and made spectral diagnosis analysis, based on whether the spectral peak of the characteristic elements contained in the measured optical elements appeared or not. Geng et al. [21] put forward to distinguish the damage state of films by using the gray and color information from the film sample image irradiated by laser, which proved that the gray value of the film sample image would increase with the increase of laser irradiation energy.
However, the above studies are all based on single type source information to evaluate the damage state of CCD. The characteristics of single-source information determine that it is easy to misjudge in complex environment, and the overall evaluation accuracy is low. With the development of multi-information fusion technology, many fusion algorithms have been put forward and successfully applied in various fields [23–25]. Various information sources [26–28] generated in the process of laser damage to CCD can be combined with information fusion algorithms to improve the evaluation accuracy.
In this paper, the damage evaluation technology of CCD based on multi-source and multi-information such as ‘cat's eye’ echo information, plasma flash spectrum information and surface image information are mainly studied, including multi-source information theory research, multi-source information acquisition system construction, multi-source and multi-information fusion method research and multi-source and multi-information fusion evaluation experiment. Experimental data analysis proves that multi-source and multi-information fusion method can get higher evaluation accuracy than single-source information.
2. Characteristics of information in laser damaging CCD
CCD operates by converting optical images into digital signals through the charge-coupling principle. Its fundamental structure consists of several layers, progressing from the surface to the interior: the microlens layer, dichroic filter layer, light-shielding aluminum film layer, SiO2 insulating layer, and Si base layer, as illustrated in Fig. 1(a). When a high-power laser irradiates the CCD, the Si substrate and the light-shielding aluminum film typically absorb the majority of the laser energy. The resulting heat not only elevates their temperatures but also conducts to the surrounding areas, leading to differential temperature effects across the various layers. If the temperature rises without reaching the material's melting point, it may induce phase transitions and increase material conductivity, particularly in the insulating layer. This can result in weak currents within the MOS structure, leading to point damage. Figure 1(b) depicts the CCD output image after point damage, with white dots indicating the damage points.
Continued temperature escalation may cause the insulating layer material to melt, possibly doping some metal or Si atoms, thereby disrupting the MOS structure. At this stage, a current source forms, leading to line damage when vertical displacement occurs, causing the entire column to be charged. Figure 1(c) illustrates the CCD output image after line damage, with the white line outlining the column output damage line.
As the molten area spreads further, more MOS structures may be destroyed, resulting in electrical connections between transfer electrodes. This hampers the CCD's charge transfer ability, leading to surface damage. The output image of the CCD after surface damage, shown in Fig. 1(d), reveals a white screen devoid of any scene information. Once in this state, the CCD loses its imaging capability entirely, and even after a period without laser irradiation, the image cannot be recovered.
When laser causes point damage and line damage to CCD, generally no obvious change can be observed from the CCD surface, but when the laser energy is large enough to cause surface damage to CCD (that is, complete damage), the roughness and surface morphology of CCD surface will change obviously, as shown in Fig. 2. Figure 2(a) and Fig. 2(b) show the CCD’s surface microscope image when undamaged and completely damaged. It can be seen that the CCD’s surface distribution is uniform and the color is regular when undamaged, but after completely damaged, the Si base layer is exposed and the color is metallic. Figure 2(c) and Fig. 2(d) are the surface profiles of undamaged and completely damaged CCD respectively. It can be seen that the surface morphology and roughness are obviously different when undamaged and completely damaged.
The changes of CCD’s surface characteristics before and after damage can be obtained in many non-contact ways, among which the changes of surface color and distribution can be obtained in real time through microscopic imaging, and the changes of surface roughness can be obtained through active laser echo detection (the echo intensity and polarization characteristics will change due to the changes of surface roughness). In addition, plasma flash will be produced during the complete damage of CCD by intense laser, and the damage state of CCD can also be evaluated by analyzing the plasma spectrum.
3. Multi-source information feature acquisition in damage process
In this paper, the experimental setup depicted in Fig. 3 is utilized to assess the outcome of laser-induced damage on CCD. The setup comprises a laser damage system, a ‘cat's eye’ echo detection system, a plasma detection system, and a microscopic imaging detection system. The laser damage system is employed to induce complete damage to the CCD using pulsed laser energy. The ‘cat's eye’ echo detection system serves for active laser echo detection of the CCD surface profile. A detection laser emits a light beam onto the CCD surface, and the resulting echo is directed by a beam splitter (BS2) towards the polarization imaging detection system, composed of a polarization beam splitter prism (PBS), CCD1, and CCD2. The microscopic imaging detection system is utilized for capturing microscopic images of the CCD surface features. Lastly, the plasma detection system is responsible for capturing the plasma flash spectrum generated when intense laser irradiates the CCD. This system includes an optical fiber spectrometer, an optical fiber coupling mirror, and an optical fiber.
3.1 Laser damaging CCD experiment
In the experiment, a 1064 nm pulse laser serves as the damaging source, characterized by a pulse width of 10 ns and adjustable pulse energy and repetition frequency. Extensive damage experiments reveal that when the energy of a single pulse remains below 5mJ, only a temporary dazzling effect occurs upon each pulse interacting with the CCD. Even after hundreds of pulses, the CCD consistently recovers to its normal imaging state, suggesting that 5mJ is insufficient to induce permanent damage. However, when the energy of a single pulse reaches 17mJ, it results in either line or surface damage, indicating a critical damage threshold. Beyond this threshold, with single pulses exceeding 17mJ or the cumulative effect of multiple pulses, complete CCD damage ensues with a 100% probability of occurrence.
To ensure a clear differentiation between the undamaged and completely damaged states, two experimental protocols are employed. In the first approach, a single pulse with 4.5mJ energy irradiates the CCD, inducing no damage or surface alteration, as depicted in Fig. 4(a). In the second experiment, a single pulse energy of 17.5mJ is applied, with five laser pulses delivered as a group (5n pulses, where n = 1,2,3,4,5), resulting in complete CCD damage for each group. Figures 4(b)-(d) illustrate images of the damage points after 10, 15, and 25 pulses, respectively.
Figure 5 presents the evident photographs of CCD damage points captured by a metallographic microscope under varying pulse conditions. Despite the CCD experiencing complete damage from each pulse group (consisting of five pulses), distinct surface morphologies emerge under different pulse conditions. Given the variability in pulse numbers during practical damage assessment, it becomes imperative to gather damage information across a spectrum of pulse configurations, facilitating a comprehensive and nuanced evaluation process.
3.2 ‘Cat's eye’ echo information acquisition
When a detected target with a certain reflectivity is near the focal plane of the optical imaging system and irradiated by a laser beam, according to the principle of optical path reversibility, an approximately parallel echo beam will be generated, and the echo signal far higher than that of other diffuse objects can be obtained near the laser emitting end. This phenomenon is called ‘cat's eye’ echo effect, and its equivalent optical path diagram is shown in Fig. 6(a). The echo detection system in Fig. 3 uses the ‘cat's eye’ echo effect to detect echo beam reflected by the damaged CCD on the focal plane.
In Fig. 3, the radiation polarized laser beam emitted by the detecting laser is reflected by the BS1 and focused by the focusing lens, reflected on the damaged CCD surface, and the reflected echo is reflected by BS1 and BS2, and then is divided into two orthogonal polarized beams by PBS, which are detected by CCD1 and CCD2 respectively. The light spot distribution images detected by CCD1 and CCD2 are shown in Fig. 6(b). After image processing and gray integration, two light intensity ${I_P}$ and ${I_S}$ with orthogonal polarization directions can be obtained.
Jones vector can be used to represent the Degree of Polarization (DP) of the echo beam, as shown in Formula (1).
Among them, ${I_P}$ and ${I_S}$ represent the intensities of parallel polarization and perpendicular polarization respectively. The total intensity of echo beam can be expressed by the sum of two orthogonal polarized light intensities: $I = {I_S} + {I_P}$.
Thus, the total light intensity information and polarization information of the echo beam can be obtained as the characteristic information of ‘cat's eye’ echo detection.
The detection laser used in the experiment is a 671 nm CW laser from China CNI company, model MRL-III, with maximum power of 200 mW and beam divergence angle of 1.5 mrad. The polarization beam splitter prism is a broadband polarization beam splitter prism (wavelength range is 620 nm-1000 nm). CCD1 and CCD2 are black-and-white area array cameras of Teledyne Dalsa Company of the United States and Sigma AF APO series telephoto zoom lens (focal length: 70mm-300 mm) is used as the imaging lens.
3.3 Plasma flash information acquisition
When intense pulsed laser is applied to the CCD surface, because the pulse action time is very short, the irradiated position of CCD surface material will be heated rapidly due to the absorption in the film, and gasification will occur before heat conduction occurs, and atoms in the vapor of CCD surface material will be excited or ionized, forming plasma flash, which may cause CCD to be damaged. In addition, if the intense pulsed laser is focused in the air, sufficient instantaneous field strength will also lead to air ionization and plasma flash. Therefore, it is impossible to judge whether the CCD is damaged only by the presence or absence of plasma flash, but it is necessary to analyze the plasma flash spectrum for further judgment.
Figure 7(a) shows the plasma spectrum generated on the surface of CCD. In the spectral range of 180-530 nm, plasma lines of primary ionization of Si atom, such as 251.75 nm, 288.42 nm and 263.22 nm, are observed, which obviously indicates that Si elements in CCD are ionized.
The layer containing Si element in CCD includes SiO2 insulating layer and Si substrate layer. In order to judge whether or not complete damage of CCD, the electron temperature must be considered. Using Si plasma spectrum to extract the plasma electron temperature can also eliminate the influence of detection distance, detection light path and background interference on the plasma spectrum.
Under the approximation that the plasma satisfies the Local Thermodynamics Equilibrium (LTE) condition, the electron temperature in the plasma can be determined by the relative strength of three spectral lines [26]. The distribution of particles in plasma at its bound energy level is Boltzmann distribution, and the relative intensity ratio ${I_{21}}$ of two spectral lines of the same ionization level can be expressed as:
Draw Boltzmann diagram with $Bol(m )$ and ${E_m}$, calculate its slope ${k_{23}}$, as shown in Formula (4), and then bring it into Formulas (2) and (3) to determine the electron temperature T, as shown in Formula (5).
According to the actual observed plasma spectrogram in Fig. 3(a), three primary ionized plasma spectral lines of 251.75 nm, 263.219 nm and 288.245 nm can be selected to calculate the electron temperature by formulas (2)–5). The calculation process is as follows: firstly, the NIST database is used to find the transition probability ${A_{mn}}$, statiical weight ${g_m}$ and upper level energy ${E_k}$ of the spectral line corresponding to Si element, as shown in Table 1. Secondly, according to formula (2), the relative intensities of the two spectral lines of 251.75 nm and 288.245 nm are calculated based on the spectral line of 263.219 nm. Thirdly, according to formula (4), the slope ${k_{23}}$ is obtained by using the upper energy levels of spectral line transitions corresponding to 251.75 nm and 288.245 nm. Finally, the plasma electron temperature t is obtained by calculating $1/{k_{23}}$.
Table 1. Si plasma parameters
In the plasma detection system depicted in Fig. 3, the optical fiber spectrometer is a UV-NIR high-resolution model BIM-6002A-09 sourced from BroLight Company in China, operating within a wavelength range of 180 nm to 530 nm. To capture the relevant plasma spectrum effectively, the optical fiber spectrometer is configured to automatically collect spectrum signals 100 ms after the emission of the damaging laser.
Figure 7(b) displays the electron temperature calculated for three distinct damage points. It's evident that the electron temperature escalates rapidly from 2800 K to 3700 K between the first and fifth pulses, eventually stabilizing around 3600 K thereafter. This trend holds consistent across all three damage points. Analyzing the curve of electron temperature variation, the electron temperatures corresponding to the first, fifth, and 25th (last) pulses respectively signify the initial, critical, and stable states. Consequently, these values are chosen as the plasma characteristic parameters for assessing the damage status of the CCD.
3.4 Surface image information acquisition
The surface image information is also important data for evaluating the damage state of CCD. Figure 8 shows the damaged CCD surface image taken by a microscope camera lens. From the figure, there are obvious differences between the damaged area and the undamaged area, and the feature information can be obtained by image processing algorithm. Among them, the microscope lens is XDC-10A continuous zoom industrial microscope lens made by Shunlihua Company, with the objective lens magnification of 1X-4.5X, the eyepiece magnification of 0.5X and the working distance of 100mm-120 mm. The camera is a near-infrared enhanced CMOS black-and-white area array camera from BASLER company, the model is acA2000-165umNIR, the detector resolution is 2048 × 2048, and the single pixel size is 5.5$\mu m$.
By comparison, three image features are extracted as evaluation indexes, namely, surface roughness information, gray variance information and color ratio information.
Among them, the surface roughness of CCD is characterized by surface texture structure, Tamura texture is a commonly used texture extraction method [28], and Tamura roughness is the most basic texture feature, which reflects the grain size in the texture. The larger the grain size, the rougher the texture image, and vice versa. The feature extraction of CCD surface roughness in the image is divided into three steps: (1) Calculating the average intensity value of pixels in the active window with the size of $2k \times 2k$ pixels in the image, (2) For each pixel, the average intensity difference between windows which are not overlapped in the vertical and horizontal directions is calculated respectively, (3) Calculate the average value of intensity difference in the whole image to get the roughness.
Gray variance reflects the average degree of gray change in the image. The greater the average degree of gray change, the clearer the image, and the smaller the average degree of gray change, the more blurred the image. Therefore, when the CCD surface is undamaged, the gray variance of the CCD surface image is large, while when the CCD surface is completely damaged, the gray variance of the CCD surface image is small. The difference between different damage states of the CCD surface is obvious, which can be used to evaluate the CCD damage state. The process [29] of extracting gray variance from CCD surface image is as follows: taking the average gray value of all pixels in the image as a reference, summing the squares of the gray values of each pixel and standardizing them.
The color ratio can also characterize the surface morphology of CCD. When the CCD surface is undamaged, there are many pixels with gray value between 100 and 150 in the CCD surface image. When the CCD surface is completely damaged, there are many pixels with gray value between 0 and 100 in the CCD surface image. The ratio of the number of pixels with gray value in the range of 0-75 is taken as the white-black ratio as the evaluation index.
Table 2 shows the difference of each feature quantity of the surface image in the state of complete damage and undamaged. Among them, groups 1-5 in Table 2 are the microscopic imaging features of the damage points shown in Figs. 5(b)-(f) respectively, and the undamaged images are the undamaged areas near the corresponding damage points. As can be seen from Table 2, there are obvious differences in surface roughness, gray variance and color ratio between the damaged area and the undamaged area.
Table 2. Surface image feature difference
3.5 Multi-source information fusion algorithm
Probabilistic Neural Networks (PNN) is adopted as the fusion algorithm of multi-source information [23,24]. This method refers to a method that uses posterior probability density to predict evaluation categories, based on Bayesian minimum risk criterion, with maximum posterior probability as its essence and radial basis function as its basic factor. Probabilistic neural network consists of input layer, hidden layer, summation layer and output layer, as shown in Fig. 9.
Among them, the input layer is to transmit the feature vector $x = [{{x_{1,}}{\; }{x_{2,}}{\; } \ldots ,{x_n}} ]$ into the network, where n represents the number of features; The hidden layer is connected with the input layer through the connection weight, and the similarity between the test point and each point in the training set is calculated, and the Gaussian function value ${\varphi _{ij}}(x )$ is output, as shown in Formula (6):
PNN method is suitable for large-scale data sets and has fast training speed, excellent performance in dealing with multi-classification problems, and has certain fault tolerance to noise and outliers.
4. Information fusion results and analysis
4.1 Data preprocessing
In order to study the evaluation method of multi-information fusion, it is necessary to preprocess the collected information features. ‘Cat's eye’ echo information includes intensity and polarization, plasma flash information includes three electron temperature characteristics (electron temperature corresponding to the first, fifth and last pulses), and surface image information includes such three characteristics as surface roughness, gray variance and color ratio. The information data are all composed of feature information and category labels, wherein the category labels are composed of completely damaged points and undamaged points,’1’is completely damaged points, and ‘2’ represents undamaged points.
4.2 Single-source information evaluation experiment and result analysis
4.2.1 ‘Cat's Eye’ echo single source information fusion experiment
PNN method is used to analyze the collected ‘cat's eye’ echo data. There are 3300 groups of data, 2310 groups are randomly selected as training sets and 990 groups as test sets to prevent uneven division, and total data are divided as five data sets 1-5 for comparison.
Figure 10(a) illustrates the data classification of dataset 1, with red dots denoting damaged data and green square points representing undamaged data. The x-axis signifies the normalized intensity of the echo beam, while the y-axis denotes the echo polarization degree. It's apparent that one-dimensional information (either intensity or polarization degree alone) struggles to effectively differentiate between damaged and undamaged points. However, the optimal classification boundary, depicted by the black dotted line, underscores the necessity for two-dimensional information encompassing intensity and polarization to achieve better discrimination between damaged and undamaged points. This principle holds true for the other four datasets as well.
Through analysis of dataset 1, the relationship between the smoothing factor σ of the Probabilistic Neural Network (PNN) and the error rate can be discerned, as depicted in Fig. 10(b). Notably, when the smoothing factor σ approaches 0.05, the error rate reaches its nadir at 11.7%. However, it's worth noting the presence of misclassified data points represented by the blue cross line in Fig. 10(a), indicative of residual misclassification errors despite the adoption of two-dimensional information. These errors likely stem from noise and measurement inaccuracies.
Refer to Table 3 for comprehensive evaluation outcomes across the five datasets. The evaluation efficacy across each dataset in the ‘Cat's Eye’ echo information fusion experiment demonstrates close consistency, with an average error rate of 10.65% and evaluation times fluctuating around 57 ms.
Table 3. Evaluation results
4.2.2 Plasma flash information evaluation experiment
Similarly, 3300 sets of plasma data are divided into five datasets labeled 1-5. Each dataset comprises three electron temperatures: those obtained at the first pulse, the fifth pulse, and the final pulse. Initial testing reveals that relying solely on one electron temperature among the three results in a classification error rate exceeding 40%. To enhance evaluation accuracy, training tests are conducted by combining the electron temperatures in pairs.
The Probabilistic Neural Network (PNN) method is employed to assess the collected plasma flash information, with the evaluation error rate depicted in Fig. 11. Within the figure, Te_15 denotes the fusion results combining the first and fifth electron temperatures, Te_1n represents the fusion results of the first and final electron temperatures, and Te_5n corresponds to the fusion results of the fifth and final electron temperatures. As observed in Fig. 11, the fusion results achieved by combining the first and final electron temperature features consistently maintain the lowest error rate across all smoothing factors. Consequently, the first and final electron temperature features are selected as the evaluation features.
Refer to Table 4 for the evaluation outcomes of the five datasets selected using Te1n. Across these datasets, the evaluation results exhibit uniformity, with error rates ranging from 5.36% to 12.5%, an average error rate of 7.3%, and a prediction time of 0.7 ms.
Table 4. Evaluation results
4.2.3 Surface image information evaluation experiment
Following evaluation, it was observed that the error rates obtained using only surface roughness, gray variance, or color ratio individually exceeded 40%. Thus, the combination of these three features was retained. Figure 12 illustrates the evaluation error rates obtained through pairwise pairing among surface roughness, gray variance, and color ratio.
As depicted in Fig. 12, the combination of roughness and gray variance exhibits the lowest error rate when σ exceeds 0.4, with an already very low error rate achieved at σ = 0.7. The evaluation outcomes for the five datasets are presented in Table 5. Across each dataset, the error rates range from 5.91% to 8.65%, yielding an average error rate of 7.17% and an evaluation time of approximately 55 ms.
Table 5. Evaluation results
4.3 Three-source information evaluation experiment and result analysis
Despite the effectiveness of employing two features, environmental noise and measurement errors continue to significantly impact the error rates, as evidenced by the data in Tables 3, 4, and 5. To further enhance evaluation accuracy, the Probabilistic Neural Network (PNN) method is utilized to fuse information from three sources: cat's eye echo, plasma flash, and surface image. Detailed experimental results are presented in Table 6. Notably, the evaluation outcomes across the five datasets of the three-source information fusion experiment exhibit similarity, with error rates ranging from 0.52% to 0.84% and prediction times ranging from 55.3 ms to 55.8 ms. Comparatively, the error rate of the three-source fusion experiment is approximately 10% lower than that of the single-source ‘cat's eye’ echo experiment, 9% lower than that of the single-source plasma flash information fusion experiment, and 6% lower than that of the single-source surface image information fusion experiment.
Table 6. Evaluation results of three-source information fusion
In summary, based on the experimental results and multi-source fusion analysis, it is evident that increasing the number of information sources enhances evaluation accuracy. The experimental findings demonstrate that the error rate of the evaluation results obtained through three-source fusion, incorporating ‘cat's eye’ echo, plasma flash, and surface image information, consistently remains below 0.85%.
5. Conclusion
This paper presents an effective approach for evaluating the damage state of CCD by leveraging a fusion method that integrates various information acquired during the process of laser-induced damage. Specifically, we constructed both a laser damage CCD experimental system and multiple information acquisition experimental systems to gather three types of crucial data: ‘cat's eye’ echo information, plasma flash information, and CCD surface image information. The ‘cat's eye’ echo information captures two key characteristics: intensity and polarization, while plasma flash information yields electron temperature data at initial and stable time points, and CCD surface image information provides insights into roughness and gray variance. Utilizing the probabilistic neural network method, we conducted fusion experiments to evaluate CCD damage based on the diverse feature information collected. The experimental results reveal a significant reduction in error rates—10%, 9%, and 6% lower, respectively—compared to single-source information evaluation results for ‘cat's eye’ echo, plasma flash, and surface image. Across multiple datasets, the error rates consistently remain below 0.85%, with evaluation times under 60 ms. These findings validate the efficacy and efficiency of the multi-source information fusion evaluation method, demonstrating its applicability for online and real-time assessment of laser-induced CCD damage.
Disclosures
The authors declare no conflicts of interest.
Data availability
Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.
References
1. X. Song, H. Chen, and Y. Xue, “Stabilization precision control methods of photoelectric aim-stabilized system,” Opt. Commun. 351(9), 115–120 (2015). [CrossRef]
2. X. Zhou, Y. Jia, Q. Zhao, et al., “Dual-rate-loop control based on disturbance observer of angular acceleration for a three-axis aerial inertially stabilized platform,” ISA Trans. 63, 288–298 (2016). [CrossRef]
3. M. Xie, P. Liu, C. Ma, et al., “Application research of high-precision laser beam pointing technology in airborne aiming pod,” Optik 183(4), 775–782 (2019). [CrossRef]
4. S. Xu, S. Wang, W. Guan, et al., “Analysis of photoelectric countermeasure technology in modern war,” Infrared 35(4), 1–6 (2014).
5. D. Abbasi-Moghadam, M. Lotfi, H. Ahmadi, et al., “Design and implementation of camera CCD readout for a remote sensing LEO satellite,” Optik 127(8), 4178–4184 (2016). [CrossRef]
6. T. Wei, R. Wang, T. Wang, et al., “Outfield experiment of semiconductor laser jamming on color CCD camera,” Optik 173(11), 185–192 (2018). [CrossRef]
7. J. Xu, S. Zhao, R. Hou, et al., “Laser-jamming analysis of combined fiber lasers to imaging CCD,” Opt. Lasers Eng. 47(7-8), 800–806 (2009). [CrossRef]
8. M. Li, G. Jin, Y. Tan, et al., “Study on the mechanism of a charge-coupled device detector irradiated by millisecond pulse laser under functional loss,” Appl. Opt. 55(6), 1257–1261 (2016). [CrossRef]
9. G. Li, “Laser-induced damages to charge coupled device detector using a high-repetition-rate and high-peak-power laser,” Opt. Laser Technol. 47, 221–227 (2013). [CrossRef]
10. W. Li, “New analysis on laser-induced damage mechanism of array CCD device,” High Power Laser Part. Beams 17, 1457–1460 (2005).
11. H. Li, X. Wang, J. Nie, et al., “Influence of pulse width on laser damage effect of CCD detector,” Infrared Laser Eng. 42, 403–406 (2013).
12. R. Gao, C. Niu, X. Li, et al., “Laser damage identification method and development status of photoelectric detectors,” Infrar. Tecchnol. 38(8), 636–642 (2016).
13. D. Qiu, “Mechanism research of pulsed-laser induced damage to CCD imaging devices,” Acta Opt. Sin. 31, 0214006 (2011). [CrossRef]
14. X. Li, C. Niu, Y. Lv, et al., “Influence of aperture diffraction on the echo power distribution of cat's eye,” Appl. Opt. 37(5), 776–782 (2016). [CrossRef]
15. M. Zhang, J. Nie, K. Sun Ke, et al., “Experimental study on laser cat's eye echo of optical imaging system during CCD damage,” Acta Photonica 48(3), 23–30 (2019).
16. B. Zhang, X. Zhang, D. Wu, et al., “Analysis of echo power of “cat's eye” target under oblique incidence,” Laser and Infrared 39(10), 1046–1050 (2009).
17. P. Lei, K. Sun, Y. Zhang, et al., “Cat's eye echo detection of damage evolution of silicon induced by 1.06um laser,” Infrared Laser Eng. 45(12), 94–100 (2016).
18. X. Qin, C. Niu, S. Chen, et al., “Study on polarization characteristics of echo scattering of cat's eye target based on micro-bin theory,” Appl. Opt. 41(5), 916–923 (2020). [CrossRef]
19. W. Hu, Y. Lv, R. Geng, et al., “Polarization imaging detection system for surface damage of photoelectric detector,” Infrared Laser Eng. 51(06), 359–367 (2022).
20. J. Ge, J. Su, L. Chen, et al., “A new method for testing laser damage threshold,” Spectr. and Spectr.l Analys. 36(5), 1296–1299 (2016).
21. S. Geng Shuai, “Research on laser film damage identification method based on image features,” Xi ‘an Univer. Technol. 2019.
22. G. Hallo, C. Lacombe, J. Néauport, et al., “Detection and tracking of laser damage sites on fused silica components by digital image correlation,” Opt. Laser Eng. 146(11), 106674 (2021). [CrossRef]
23. J. Li, Y. Fan, and B. He, “Research on tool wear based on improved particle swarm optimization and PNN neural network based on parameter strategy,” Machine Tool Hydr. 49(3), 75–80 (2021).
24. Q. Zhang, W. Yu, and C. Wang, “Study on the identification of the cutting teeth wear degree of roadheader based on PNN neural network,” Coal Science Technol. 47(6), 37–44 (2019).
25. Y. Jiang, G. Yi, R. Huang, et al., “Transformer detection and evaluation technology based on multi-source information fusion,” J. Shanghai Electr. Power Univer. 36(5), 481–485 (2020).
26. M. Su, Q. Min, Q. Zhang, et al., “Spectral analysis of laser-generated Si plasma,” J. Northwest Normal Univer. 50(1), 39–42 (2014).
27. L. Liu, L. Qu, Y. Tan, et al., “Plasma spectrum analysis of monocrystalline silicon irradiated by pulsed laser,” High Power Laser Part. Beams 22(8), 1815–1818 (2010). [CrossRef]
28. Y. Liu, “Research on the application of workpiece surface roughness detection based on laser speckle image,” Shenyang Univer. Aeron. Astron. (2019).
29. Z. Cai, T. Chen, C. Ceng, et al., “Research on fingerprint image segmentation algorithm combining direction and gray information,” J. Wuhan Univer. Technol. 39(4), 94–98 (2017).
