The paper aims to propose an image deblurring method based on deep reinforcement learning to effectively remove multiple image blurs and improve image quality. GoPro and DIV2K datasets were used for experiments. The peak signal-to-noise ratio (PSNR) and structural similarity index (SSIM) were used as objective evaluation indicators. The high-dimensional feature of fuzzy image was obtained by convolutional neural network. The deblurring framework was established by deep reinforcement learning combined with a variety of CNN deblurring tools. The peak signal-to-noise ratio (PSNR) was used as the training reward evaluation function to select the optimal restoration strategy and gradually restore the fuzzy image. Through training and testing, compared with the existing mainstream algorithm, the method presented in this paper had a better subjective visual effect; and the PSNR value and SSIM value had better performance. The experimental results show that the method in this paper can effectively solve the problem of Gaussian blur and motion blur of image, and obtain good visual effects. It has certain reference value in image deblurring.