Image segmentation is the process of extracting objects of interest from an image, playing a vital role in automated image analysis. In this study, we aim to enhance image segmentation by developing a modified version of the fuzzy K-means algorithm and introducing pre- and post-processing techniques. While hard K-means clustering is a commonly used algorithm for clustering, there are also Mathematical Morphology (MM)-based segmentation techniques such as Waterfall, Watershed, and P algorithms. Among the fuzzy clustering algorithms, the fuzzy K-means algorithm is widely recognized. Our approach involves implementing a pre-processing method for hard K-means, a post-processing method for the P algorithm, and an improved version of the fuzzy K-means algorithm. These implementations contribute to achieving superior image segmentation compared to existing algorithms. To improve the accuracy of the segmentation, we employ the grab-cut algorithm as a pre-processing step. This algorithm extracts the foreground from the input image, which is then used as input for the k-means clustering algorithm. Additionally, our post-processing method effectively eliminates noise-related features from the output of the P algorithm. Furthermore, the modified fuzzy K-means algorithm progressively eliminates the least suitable cluster through a series of iterations until a predetermined number of clusters is obtained. The results demonstrate the superiority of the modified fuzzy C-means algorithm, outperforming the hard K-means algorithm in 88% of the tested images by producing better clusters. The post-processing of the P algorithm also yields marginally improved results, successfully eliminating noisy features from all tested images. However, the pre-processed K-means algorithm exhibits mixed performance due to the use of a simplistic foreground mask for extraction. Nevertheless, it still outperforms the K-means algorithm in 80% of the tested images.