1. Introduction
In recent years, with the increasing emphasis on renewable energy, solar energy has become one of the most important sources of clean energy. Solar panels, as the key components of solar power generation systems, play a crucial role in converting solar energy into electrical energy. However, in some areas with severe air pollution or in foggy weather conditions, the performance of solar panels may be affected, and the captured images may be blurred or degraded due to the presence of haze or fog. This not only affects the visual inspection and monitoring of solar panel conditions but also may lead to inaccurate analysis and evaluation of solar panel performance. Therefore, it is of great significance to study effective image defogging methods for solar panel images.
1.1 Background and Motivation
The wide application of solar panels in various fields has brought about a large amount of image data. In order to ensure the normal operation and efficient power generation of solar power generation systems, it is necessary to monitor and manage the working status of solar panels in real time. However, foggy weather often occurs in many regions, which reduces the visibility of solar panel images and makes it difficult to accurately judge the surface condition, damage, and performance degradation of solar panels. Therefore, developing an efficient and accurate image defogging algorithm for solar panel images is an urgent problem to be solved.
1.2 Research Objectives
The main objective of this study is to propose an improved MSRCR (Multi-Scale Retinex with Color Restoration) algorithm for defogging solar panel images, aiming to improve the quality of foggy solar panel images, enhance the contrast and color saturation of the images, and restore the details and true colors of the images as much as possible. At the same time, through experimental comparison and analysis with other traditional defogging algorithms, the superiority and effectiveness of the proposed algorithm are verified.
1.3 Significance of the Study
This research has important theoretical and practical significance. On the one hand, it enriches and develops the theory and method of image defogging technology, providing a new idea and solution for improving the quality of foggy images. On the other hand, it can be applied to the monitoring and management of solar panel systems in actual production and life, helping to improve the operation and maintenance efficiency of solar panel systems, reduce energy losses caused by image quality problems, and promote the development and utilization of solar energy.
2. Related Work
In the field of image defogging, many scholars have conducted extensive research and proposed a variety of algorithms. The following is a review and analysis of some representative algorithms.
2.1 Traditional Defogging Algorithms
- Dark Channel Prior Algorithm: This algorithm is based on the statistical characteristics of the dark channel in the image. By estimating the atmospheric light and transmission rate, the fog in the image is removed. However, it may cause color distortion and edge over-enhancement in some cases, and the computational complexity is relatively high.
- Histogram Equalization Algorithm: This algorithm improves the contrast of the image by equalizing the histogram of the image. However, it may lead to the loss of some details in the image and the generation of false contours.
- Wavelet Transform Algorithm: This algorithm decomposes the image into different frequency bands using wavelet transform and then enhances the details of the image by processing the high-frequency components. However, the selection of wavelet basis functions and decomposition levels has a certain impact on the defogging effect, and it is also sensitive to noise.
2.2 Retinex-Based Defogging Algorithms
Retinex theory is a classic image enhancement theory, which believes that the image can be decomposed into illumination and reflection components. Based on this theory, many Retinex-based defogging algorithms have been proposed.
- SSR (Single Scale Retinex) Algorithm: This algorithm performs Retinex transformation on the image at a single scale to enhance the contrast and details of the image. However, it may not be able to effectively handle images with complex lighting conditions and strong fog.
- MSR (Multi-Scale Retinex) Algorithm: In order to overcome the limitations of the SSR algorithm, the MSR algorithm performs Retinex transformation on the image at multiple scales and then combines the results of different scales to obtain a better defogging effect. However, the MSR algorithm also has some problems, such as color distortion and relatively high computational complexity.
- MSRCR (Multi-Scale Retinex with Color Restoration) Algorithm: The MSRCR algorithm further improves the MSR algorithm by introducing a color restoration step to improve the color reproduction ability of the image. However, the parameter selection of the MSRCR algorithm is relatively difficult, and in the case of strong fog, its effect may not be satisfactory.
2.3 Deep Learning-Based Defogging Algorithms
With the development of deep learning technology, some scholars have begun to apply deep learning methods to image defogging and achieved good results.
- CNN-Based Defogging Algorithms: Convolutional neural networks (CNNs) are used to learn the mapping relationship between foggy images and clear images, and then the defogging of foggy images is realized. These algorithms can automatically learn the features of foggy images and have strong adaptability and generalization ability. However, they require a large amount of training data and computational resources, and the training process is time-consuming.
- GAN-Based Defogging Algorithms: Generative adversarial networks (GANs) are used to generate clear images from foggy images. By training the generator and discriminator in the GAN, the quality of the generated clear images is continuously improved. However, GAN-based defogging algorithms are also faced with problems such as unstable training and difficult control of the generated results.
The following table summarizes the characteristics of the above algorithms:
| Algorithm | Advantages | Disadvantages |
|---|---|---|
| Dark Channel Prior Algorithm | Good defogging effect in some cases | Color distortion, high computational complexity |
| Histogram Equalization Algorithm | Simple and fast, can improve contrast | Loss of details, false contours |
| Wavelet Transform Algorithm | Can enhance image details | Sensitive to noise, parameter selection affects effect |
| SSR Algorithm | Simple implementation | Limited defogging ability |
| MSR Algorithm | Better defogging effect than SSR | Color distortion, high computational complexity |
| MSRCR Algorithm | Improves color reproduction | Parameter selection difficult, poor effect in strong fog |
| CNN-Based Defogging Algorithms | Strong adaptability and generalization ability | Require large amount of data and resources, time-consuming training |
| GAN-Based Defogging Algorithms | Can generate clear images | Unstable training, difficult to control results |
3. Methodology
In this study, an improved MSRCR algorithm is proposed for defogging solar panel images. The following is a detailed description of the algorithm.
3.1 Bilateral Filtering
Bilateral filtering is an effective image filtering method that can smooth the image while preserving the edge information. The output of the bilateral filter at a pixel position (x,y) is given by the following formula:
where G(x,y) is the output pixel value, I(x+di,y) is the value of the neighboring pixel, ωi(di) is the spatial weight (usually a Gaussian function), and ωr(I(x+di,y)-I(x,y)) is the intensity weight. The normalization factor ωp ensures that the sum of the weights is 1.
3.2 MSRCR Algorithm
The MSRCR algorithm combines the Retinex theory and multi-scale processing to enhance the contrast and color restoration of the image.
Single-Scale Retinex Transform: For a given image I(x,y), the single-scale Retinex transform is defined as:
where di is the pixel offset in the spatial neighborhood and N is the number of pixels in the neighborhood.
Multi-Scale Retinex: To handle different scales of information, the MSRCR algorithm performs multi-scale decomposition on the image. For each scale, the single-scale Retinex transform is applied, and the final enhanced image is obtained by combining the results of different scales:
where Rs(x,y) is the transform at the s-th scale, S is the total number of scales, and ωs is the weight.
Color Restoration: To preserve the color information of the image, the MSRCR algorithm introduces a color restoration step. This involves adjusting the image in the color channels:
where C is the color restoration parameter.
3.3 Integration of Bilateral Filtering into MSRCR
The improved MSRCR algorithm integrates bilateral filtering into the Retinex transform to adjust the contrast. Specifically, the output of the bilateral filter G(x,y) is substituted into the original MSRCR formula:
By using appropriate mathematical operations, such as applying the exponential function to eliminate the logarithmic terms, the formula can be simplified as:
where α and β are optimization parameters that can be determined using the gradient descent method in experiments.
3.4 Bilateral Filtering Improved MSRCR Model
The overall process of the bilateral filtering improved MSRCR model is as follows:
- The original image is input into the multi-scale Retinex transform for scale decomposition and contrast enhancement.
- The bilateral filter is applied to the Retinex-transformed image components at each scale to optimize them, considering the spatial and intensity relationships, to improve the contrast and preserve the image details.
- The optimized image components at different scales are combined into the final enhanced image, and the defogging operation is performed to make the image have a clear appearance.
In this section, the proposed algorithm is evaluated through a series of experiments, and the results are compared with those of other traditional defogging algorithms.
4.1 Experimental Setup
The experimental data used in this study are solar panel images collected in a certain area of Shaanxi. The experimental environment is a PC with Pycharm 2022 installed. The performance of the algorithm is evaluated using the following metrics:
- Information Entropy: Measures the amount of information in the image. A higher information entropy indicates that the image contains more information.
- Average Gradient: Describes the texture change and detail contrast information of the image. A higher average gradient indicates that the image has more detailed texture and better contrast.
- Spatial Frequency: Represents the frequency components of the image in the spatial domain. A higher spatial frequency indicates that the image contains more detailed information, such as texture and edges.
4.2 Experimental Results
The experimental results of different algorithms on solar panel.
It can be seen that the original image has low clarity and contrast, and it is difficult to judge the actual working status of the solar panel. The bilateral filtering algorithm significantly improves the contrast of the image, but the clarity does not change much, and the image has high exposure, resulting in a whitish overall color and poor visual effect. The Retinex and MSRCR algorithms improve the clarity and contrast compared with the original image and the bilateral filtering algorithm, and the overall color of the image is also improved. However, they have certain color distortion and noise problems. The proposed algorithm in this paper has no noise problem and color deviation, and the processing of details is better, closer to the real scene, and the defogging effect is better.
The quantitative evaluation results of different algorithms are shown in Table 2.
| Algorithm | Information Entropy | Average Gradient | Spatial Frequency |
|---|---|---|---|
| Original Image | 5.266 | 21.232 | 4789.659 |
| Bilateral Filtering | 6.862 | 45.439 | 10626.821 |
| Retinex | 7.023 | 50.626 | 11243.367 |
| MSRCR | 7.221 | 52.325 | 12003.582 |
| Improved MSRCR | 7.630 | 57.406 | 12619.470 |
It can be seen from the table that the information entropy, average gradient, and spatial frequency of the proposed algorithm are significantly higher than those of the other three algorithms, indicating that the proposed algorithm can effectively improve the quality of solar panel images with fog and has better defogging performance.
4.3 Analysis and Discussion
The experimental results show that the proposed bilateral filtering improved MSRCR algorithm has better performance than traditional defogging algorithms. The main reasons are as follows:
- The bilateral filtering can effectively remove noise while preserving the edge information of the image, which helps to improve the clarity and detail of the image.
- By integrating bilateral filtering into the MSRCR algorithm, the contrast and color saturation of the image can be better adjusted, and the color distortion problem of the traditional MSRCR algorithm can be reduced.
- The multi-scale Retinex processing in the MSRCR algorithm can capture information at different scales, which is beneficial to enhancing the overall quality of the image.
However, the proposed algorithm also has some limitations. For example, in extremely severe foggy conditions, the defogging effect may still not be ideal. In future research, we will further explore ways to improve the performance of the algorithm in extreme cases.
5. Conclusion and Future Work
In this study, an improved MSRCR algorithm based on bilateral filtering is proposed for defogging solar panel images. The algorithm combines the advantages of bilateral filtering and MSRCR algorithm to effectively improve the contrast and color saturation of the image, while reducing noise and color distortion. Through experimental comparison and analysis with traditional algorithms, the superiority and effectiveness of the proposed algorithm are verified.
5.1 Research Achievements
- A new image defogging algorithm for solar panel images is proposed, which provides a new solution for improving the quality of foggy solar panel images.
- The proposed algorithm can effectively remove the fog in the image, enhance the details and true colors of the image, and is closer to the actual situation of the solar panel.
- The experimental results show that the proposed algorithm has better performance than traditional algorithms in terms of information entropy, average gradient, and spatial frequency, which verifies the effectiveness and superiority of the algorithm.
5.2 Future Research Directions
- Explore the application of deep learning technology in solar panel image defogging to further improve the defogging effect and generalization ability of the algorithm.
- Study the impact of different weather conditions and solar panel surface conditions on the defogging algorithm and develop more adaptive algorithms.
- Optimize the algorithm to reduce the computational complexity and improve the processing speed of the algorithm, so that it can be applied to real-time monitoring and diagnosis systems of solar panel images.
- Combine image defogging with other image processing techniques, such as image segmentation and object recognition, to provide more comprehensive and accurate information for solar panel monitoring and management.
In summary, this research has made certain contributions to the field of solar panel image defogging, but there is still room for further improvement and development. We will continue to work hard to promote the application and development of solar energy technology.