Image segmentation is computer vision and analysis of images. However, several segmentations couldn’t be manageable due to high noise levels. It is tough to deal with semi-supervised fuzzy clustering since it is not robust, and reduction needs to be improved. Many researchers have developed in this cluster. Our research focuses on adding noise, reducing noise, and applying our techniques to get detailed, clear images. Now, we introduce the robust approach kernel method based on semi-supervised fuzzy clustering with Kullback-Leibler divergence. It incorporates Kullback-Leibler divergence, and semi-supervised clustering is of essential sig-nificance. The primary benefit of robust semi-supervised fuzzy clustering is that it manages uncertainty data and kernel distance measures to capture the similarity between data points, enhancing segmentation performance. We use various strategies to introduce noise and minimise it compared to kernel robustness semi-supervised fuzzy clustering with Kullback-Leibler divergence. The effectiveness of the recommended technique was evaluated using synthetic datasets, the publicly accessible simulated human brain Magnetic Resonance Imaging dataset, and the BraTS2020 medical imaging and lung computed tomography scans. The experimental findings show that the suggested method outperforms existing algorithms in peak signal noise ratio, accuracy, precision, F1 score, and Jaccard index.