Reseaching of color distortion effects when applying adaptive histogram equalization methods to video frames from mines

Introduction: This article examines methods of digital processing of video information obtained in mines. Video frames of such information are captured under low lighting conditions, often at short distances from the lens to the object, with light sources entering the shooting area, and therefore have features that require special attention. Since implementing a video surveillance system in the mining industry is quite resource-intensive, replacing it with newer technical vision systems is not always possible, making the use of video analytics advisable. The purpose of the research: the visual quality of video frames obtained in mines in order to extract the maximum amount of information from them. Methods: Preference is given to adaptive processing methods that account for the individual characteristics of images. The authors investigate the application of adaptive histogram equalization methods, including both Adaptive Histogram Equalization (AHE) and Contrast Limited Adaptive Histogram Equalization (CLAHE), to video frames, considering the presence of dark and overexposed areas, blurred regions, and the low average brightness level of the entire image. Experimental Studies: The paper separately examines algorithms for the RGB color space, involving component-wise digital processing, and the HSV color space, where processing is performed only on the V channel. Special attention is paid to the shapes of histograms in adaptive equalization. Shapes such as uniform, bell-shaped, and exponential are considered. A parameter such as the clip limit for brightness is also taken into account, and dependencies for its calculation are proposed for both RGB and HSV formats. Results: For the methods applied, quantitative characteristics such as the average brightness level of the entire video frame and separately for color components (Tenengrad) are calculated, and a comparison is made with the visual quality of the video frames. The characteristics are calculated for various clip limits and various histogram equalization shapes. Conclusion: Conclusions are drawn and recommendations are developed regarding color spaces for processing and methods for calculating the contrast limit for enhancements. Following these recommendations will minimize color distortion effects when applying histogram equalization methods to video frames from mines and tunnels.