Dual generative adversarial network for motion blur restoration and operation method thereof

Abstract

The invention provides a dual generative adversarial network for motion blur restoration. The method comprises a data set clear domain, a data set fuzzy domain, a coupled first generative adversarialnetwork and a coupled second generative adversarial network. The first generative adversarial network comprises an original generator GA and a corresponding original discriminator DA, and the second generative adversarial network comprises a dual generator GB and a corresponding dual discriminator DB. According to the method, the problem that a deblurred data set is difficult to obtain is solved,only enough blurred images and clear images need to be obtained, two identical generative adversarial networks are used for forming a dual adversarial network, and training on a non-resultant data setcan be achieved.

Description

Technical field [0001] The present invention relates to the technical field of image processing, in particular to a dual generation confrontation network for motion blur restoration and an operating method thereof. Background technique [0002] Today, when electronic devices are widely used, photos have become an important way to record life, and surveillance has become a favorable means to find suspects. In the process of image acquisition, the image quality is often affected by some factors, such as unfocused photography equipment, circuit noise, camera shake, and subject movement. Therefore, image blur is divided into many types, of which the most common and most difficult to deal with is motion blur. Camera movement and target object movement are the two main causes of motion blur. Intelligent monitoring and autonomous driving on the street require effective deblurring algorithms to remove motion blur and increase the recognition rate. Therefore, image deblurring is a part...

AI Technical Summary

Problems solved by technology

[0005] The present invention solves the problems of difficult acquisition of deblurring data sets and low practicability of deblurring algorithms, and proposes a dual generative confrontation network for motion blur restoration and its operation method, which is trained on non-paired deblurring data sets and solves the problem of To solve the difficult problem of obtaining blurred data sets, you only need to obtain enough blurred images and clear images respectively, and use two identical generative confrontation networks to form a double confrontation network, so that it can be trained on non-synthetic data sets

Images