Enhancing Low-Light Sports Motion Images with Improved Bilateral Filtering and Auto MSRCR

Abstract

Low-light sports images are very common in night sports games recording, but many details of these images are difficult to analyze. Therefore, this paper proposes a sports image enhancement method based on improved bilateral filtering to solve the problem of image blur of night sports games recording and promote the efficiency of physical education. Firstly, in the HSV color space of the original image, the MSR algorithm is applied to the V component using bilateral filtering for brightness enhancement, preserving the original color information while improving image brightness. Next, the CLAHE algorithm is employed in the LAB color space to enhance details in the initially enhanced brightness image, resulting in a more detailed image. In order to create the enhanced low-light image, the detail-enhanced image is combined with the original low-light image using the Auto MSRCR algorithm and then weighted fusion is carried out. In the end, Wiener filtering is used to process the motion blur information and produce the final processed image. The modified images are compared with the MSR, MSRCR, CLAHE, and improved GAMMA algorithms using evaluation measures such UCIQE, AG, SD, and IE to assess the algorithm's performance. The method achieves significant improvements in both visual quality and objective metrics, such as UCIQE, compared to state-of-the-art methods like MSRCR and CLAHE. Specifically, our method improves the UCIQE score by 65.24%, demonstrating superior preservation of edge details and color balance. We also show that our approach outperforms MSRCR by 35% in reducing halo artifacts and over-enhancement. These results are validated on the LOL dataset, which includes various motion-blur scenarios in sports images.