Underwater cameras face difficulties such as light attenuation, low contrast, blurring, and noise, affecting image quality and object appearance. This article discusses the importance of underwater image processing for counting fish images in challenging environments. The proposed workflow involves image collection, preprocessing with noise reduction filters, and segmentation to distinguish fish and other aquatic objects using Otsu segmentation followed by blob analysis for fish counting. This work implements a hybrid noise reduction filter having better performance in terms of image quality metrics PSNR and SSIM of 38.31 and 0.97, respectively. Also, the proposed work has achieved a fish-counting accuracy of 82.40%. The approach aims to improve visibility and accuracy in underwater images, supporting marine exploration, ecological research, and environmental conservation. Advanced underwater image processing can enhance our understanding of the underwater world and contribute to marine ecosystem preservation.