Research on Restoration Algorithm of Two-dimensional Degraded Image Based on Deep Learning

Abstract

The recognition and application of two dimensional code has been seriously affected by the image degradation produced during the process of acquisition and transmission of it. In order to fix that, analysis of the traditional image processing technology has revealed that adopting the advanced image restoration algorithm which is based on deep learning, building restoration model of two dimensional code by using the multilayer neural network, adding binary loss constraint on the basis of L1 norm, constructing the non-linear image degradation and restoration process can solve the too complicated structure, easily affected by noise and poor robustness problems caused of traditional method. The experimental result shows that this method has a good restoration effect on the degraded image of two-dimensional code, and also improves the recognition efficiency while shortens the network training time, more over it meets the needs of practical application of two-dimensional code.