This paper proposes a deep learning method for single image super-resolution (SR) using a deep convolutional neural network (CNN). The method directly learns an end-to-end mapping between low- and high-resolution images, achieving state-of-the-art restoration quality and fast speed. Unlike traditional sparse-coding-based methods, the proposed approach jointly optimizes all layers of the network, eliminating the need for explicit dictionary learning. The network is lightweight and can process three color channels simultaneously, leading to improved performance. The method is trained using a large dataset and shows superior performance compared to existing methods in terms of PSNR, SSIM, and other evaluation metrics. The network is also efficient, with fast inference speed suitable for real-time applications. The paper also explores different network structures and parameter settings to balance performance and speed. Experiments show that the proposed method outperforms existing state-of-the-art methods in both quantitative and qualitative evaluations. The method is flexible and can be applied to color images by processing all three channels simultaneously. The results demonstrate that deep learning is effective for classical computer vision tasks like super-resolution, achieving high quality and speed. The paper also discusses the relationship between the proposed method and traditional sparse-coding-based methods, showing that the former can be viewed as a deep CNN. The method is further extended to handle color images, achieving better performance than single-channel approaches. Overall, the proposed method provides a novel and effective solution for image super-resolution.This paper proposes a deep learning method for single image super-resolution (SR) using a deep convolutional neural network (CNN). The method directly learns an end-to-end mapping between low- and high-resolution images, achieving state-of-the-art restoration quality and fast speed. Unlike traditional sparse-coding-based methods, the proposed approach jointly optimizes all layers of the network, eliminating the need for explicit dictionary learning. The network is lightweight and can process three color channels simultaneously, leading to improved performance. The method is trained using a large dataset and shows superior performance compared to existing methods in terms of PSNR, SSIM, and other evaluation metrics. The network is also efficient, with fast inference speed suitable for real-time applications. The paper also explores different network structures and parameter settings to balance performance and speed. Experiments show that the proposed method outperforms existing state-of-the-art methods in both quantitative and qualitative evaluations. The method is flexible and can be applied to color images by processing all three channels simultaneously. The results demonstrate that deep learning is effective for classical computer vision tasks like super-resolution, achieving high quality and speed. The paper also discusses the relationship between the proposed method and traditional sparse-coding-based methods, showing that the former can be viewed as a deep CNN. The method is further extended to handle color images, achieving better performance than single-channel approaches. Overall, the proposed method provides a novel and effective solution for image super-resolution.