Low-light image enhancement is an essential requirement for various applications in different fields such as image processing and computer vision. In this paper, a new approach to enhance the low-light images has been proposed. The proposed model consists of two main steps: preprocessing of the input images and the CNN model based on residual learning with skip connections. The architecture of the suggested CNN model contains three main residual blocks, each block has groups of convolution layers with activation functions and add operations. In the suggested model, the convolution layers will process the image, and some of these layers’ output will be stored in the add operation in order to forward them to the next convolution groups in the networks as input while skipping some connections, this process represents the “residual learning”. The proposed approach has been implemented on four datasets (LOL, MEF, NPE, and DICM) and compared with other state-of-the-art methods. The qualitative and quantitative results indicate excellent visual quality enhancement, improving lighting while restoring the actual colour and details of the image. And attains superior results in quality evaluation metrics, where PSNR was 34.86 in the NPE dataset, SSIM was 0.96 in the LOL dataset and NIQE was 3.18 in the MEF dataset.