Image deblurring algorithm based on deep convolution neural network
Chuanhe Huang Wentao ZHANG · Solid State Technology · 2021
Purpose to effectively remove multiple kinds of image blurring and improve image quality, an image deblurring method based on deep reinforcement learning and an image inpainting algorithm based on convolutional neural network. The high-dimensional characteristics of blurred images are obtained through convolutional neural network, and a deblurring framework is constructed through deep reinforcement learning integrated with various kinds of CNN deblurring tools, so as to select the optimal inpainting strategy and restore the blurred images step by step. The effectiveness of the above image inpainting enhancement method designed and its superiority compared with other traditional methods of the same kind are verified through subjective and objective contrast experiments