Enhancement of Mine Images through Reflectance Estimation of V Channel Using Retinex Theory

Changlin Wu,Dandan Wang,Kaifeng Huang,Long Wu
DOI: https://doi.org/10.3390/pr12061067
IF: 3.5
2024-05-24
Processes
Abstract:The dim lighting and excessive dust in underground mines often result in uneven illumination, blurriness, and loss of detail in surveillance images, which hinders subsequent intelligent image recognition. To address the limitations of the existing image enhancement algorithms in terms of generalization and accuracy, this paper proposes an unsupervised method for enhancing mine images in the hue–saturation–value (HSV) color space. Inspired by the HSV color space, the method first converts RGB images to the HSV space and integrates Retinex theory into the brightness (V channel). Additionally, a random perturbation technique is designed for the brightness. Within the same scene, a U-Net-based reflectance estimation network is constructed by enforcing consistency between the original reflectance and the perturbed reflectance, incorporating ResNeSt blocks and a multi-scale channel pixel attention module to improve accuracy. Finally, an enhanced image is obtained by recombining the original hue (H channel), brightness, and saturation (S channel), and converting back to the RGB space. Importantly, this image enhancement algorithm does not require any normally illuminated images during training. Extensive experiments demonstrated that the proposed method outperformed most existing unsupervised low-light image enhancement methods, qualitatively and quantitatively, achieving a competitive performance comparable to many supervised methods. Specifically, our method achieved the highest PSNR value of 22.18, indicating significant improvements compared to the other methods, and surpassing the second-best WCDM method by 10.3%. In terms of SSIM, our method also performed exceptionally well, achieving a value of 0.807, surpassing all other methods, and improving upon the second-place WCDM method by 19.5%. These results demonstrate that our proposed method significantly enhanced image quality and similarity, far exceeding the performance of the other algorithms.
engineering, chemical
What problem does this paper attempt to address?
The paper attempts to address the issues in underground mining environments where monitoring images suffer from uneven lighting, blurriness, and loss of details due to dim lighting and excessive dust. These problems hinder subsequent intelligent image recognition. Existing image enhancement algorithms have limitations in terms of generalization ability and accuracy. Therefore, this paper proposes an unsupervised method for mine image enhancement in the HSV color space, aiming to improve image quality and similarity, surpassing the performance of existing algorithms. Specifically, the proposed method in the paper includes the following key points: 1. **Combining Retinex Theory with HSV Color Space**: First, the RGB image is converted to the HSV color space, and Retinex theory is integrated into the brightness (V channel). A U-Net-based reflectance estimation network is designed to enhance the brightness of the image. 2. **Improved U-Net Model**: The model introduces ResNeSt blocks and a multi-scale channel pixel attention mechanism. Through operations such as convolution, encoding downsampling, feature copying, and upsampling, the V channel is enhanced. This effectively extracts image features and applies attention modules in the decoder to retain important information, resulting in excellent enhancement effects. 3. **Design of Four Loss Functions**: To address the challenges of low-light image enhancement in underground coal mines, four loss functions are designed: exposure loss function, reflectance consistency loss function, spatial structure loss function, and illumination smoothness loss function. These comprehensively consider the exposure level, reflectance consistency, spatial structure details, and illumination smoothness of the image, achieving overall optimization of the model and improving the performance of low-light image enhancement. 4. **Random Brightness Disturbance Method**: A random brightness disturbance method based on a power function is introduced. By introducing randomness during training, it simulates the reflection changes under different lighting conditions in actual scenarios, enhancing the robustness and generalization ability of the model. Through these innovations, the proposed method in the paper significantly outperforms existing unsupervised low-light image enhancement methods on multiple metrics and can even rival supervised methods in some aspects, demonstrating its potential in practical applications.