Adaptive Decision-Based Fuzzy Filtering for Image Denoising in a Spark–MongoDB Framework

Image segmentation is an important step in image processing, dividing an image into meaningful regions for tasks namely object recognition, feature extraction, and decision-making in applications like satellite imaging, medical diagnostics, and computer vision. However, segmentation accuracy heavily depends on input image quality, which is often degraded by noise. Image denoising is therefore necessary to improve image quality and keep important features. Conventional denoising methods often struggle to balance noise reduction and edge preservation, resulting to loss of critical information. This paper proposes Adaptive Decision-Based Fuzzy Membership Median Filtering (ADBFMMF) for effective image denoising. The approach includes adaptive noise identification, fuzzy membership weighting, and fuzzy-weighted median filtering to reduce noise while preserving image details. The method leverages MongoDB for scalable image storage and Apache Spark for distributed, high-speed denoising. Tests on benchmark and MRI images demonstrate superior performance over conventional filters. Performance metrics such as Mean Squared Error (MSE), Peak Signal-to-Noise Ratio (PSNR), Structural Similarity Index (SSIM), and Feature Similarity Index (FSIM) show that ADBFMMF effectively balances detail preservation and noise reduction. The results suggest that it could be useful for medical imaging, improving diagnostic accuracy and practical application in clinical settings.