To resize an image, it’s critical to maintain its geometric shape as well as its key content. A potent method that helps achieve this objective is seam carving, which guarantees little to no content loss and preserves essential components. An ideal connected path of pixels within a single image, either from left to right or from top to bottom, is referred to as a seam. The participating pixels are chosen based on their energy. In this paper, we present a novel approach to Reinforcement Learning (Q-learning) based on Alpha and special points for better content preservation, particularly in situations where other seam carving methods cause significant damage to objects in the image, leading to content distortion. Sometimes the presence of low-energy pixels inside one or a few items even results in the loss of critical content. Unlike conventional dynamic programming-based seam carving algorithms, our method uses Q-learning to make decisions about seam removal informed by learned policies. Our suggested solution uses a Q-learning agent trained to identify pixels with the lowest energy, creating an optimized seam to reduce artifacts and content loss. Through interactions with the image environment, the agent’s Q-values are obtained, which help with the seam creation process by selecting pixels based on their low energy levels. To improve resilience and help maintain important content in the image, we apply Alpha and special points. The outcomes of our experiments, conducted on a variety of image types, demonstrate the efficacy and efficiency of the strategy we’ve suggested. This work introduces a novel seam carving method based on Reinforcement Learning (Q-learning) to the domains of computer vision and image processing.