Underwater images often suffer from severe color distortion, low contrast, and reduced visibility, motivating the widespread use of image enhancement as a preprocessing step for downstream computer vision tasks. However, recent studies have questioned whether enhancement actually improves object detection performance. In this work, we conduct a comprehensive and rigorous evaluation of nine state-of-the-art enhancement methods and their interactions with modern object detectors. We propose a unified evaluation framework that integrates (1) a distribution-level quality assessment using a composite quality index (Q-index), (2) a fine-grained per-image detection protocol based on COCO-style mAP, and (3) a mixed-set upper-bound analysis that quantifies the theoretical performance achievable through ideal selective enhancement. Our findings reveal that traditional image quality metrics do not reliably predict detection performance, and that dataset-level conclusions often overlook substantial image-level variability. Through per-image evaluation, we identify numerous cases in which enhancement significantly improves detection accuracy—primarily for low-quality inputs—while also demonstrating conditions under which enhancement degrades performance. The mixed-set analysis shows that selective enhancement can yield substantial gains over both original and fully enhanced datasets, establishing a new direction for designing enhancement models optimized for downstream vision tasks. This study provides the most comprehensive evidence to date that underwater image enhancement can be beneficial for object detection when evaluated at the appropriate granularity and guided by informed selection strategies. The data generated and code developed are publicly available.

Digital images are usually suffering from noises that affect the appearance of the digital image, and this affected appearance could result in disrupting the details, edges, components, etc., and to remove these noises, we have to use one or more filtering techniques or methods. Noise is an unwanted signal that affects digital images which make the digital image has differences compared to the original image or even the real scene. Filtration is a technique used to do modifications to an image to make it more similar to the original image or scene. In this paper, we will use three methods maximum, median, and minimum filtering which are types of rank filtering. Rank filtering is a non-linear technique that uses a kernel that is a window with an odd size and has a minimum size of three, passes through all the pixels in the image and does ascend ordering, and then chooses a pixel within the order kernel to replace the center value of the kernel and reflect it to the direct equivalent position inside the filtered image. For choosing a value from the ordered kernel, a lot of rank filtering techniques appeared, and we will discuss in this paper three types, the minimum filter will choose the minimum value from the ascending ordered kernel, and the maximum filter will choose the maximum value and finally, the median filter will choose the middle value. This paper will compare these rank filters by using MATLAB, which is a scientific and engineering application, and there will be a function for removing noises for the mentioned three rank filters besides determining the performance for each type of rank filter and comparing them to find out the best type of filtering between rank filters mentioned. In this paper, the noise applied to the digital image is salt and pepper noise which is the most famous noise resulting usually from weather effects. The result of this paper is proving that median filtering has the best performance of removing salt and pepper noise compared with maximum and minimum filtering, and it is proved by using subjective and objective notice.