GAN (Generative Adversarial Network) based motion blur removing method of image

Abstract

The invention discloses a GAN based motion blur removing method of an image and a motion blur removing GAN model for the method. The method comprises that the GAN model is designed; the model is trained; and in an application phase, the GAN model comprises a generator and a discriminator, the generator optimizes parameters continuously so that a generated image approaches distribution of a clear image, the discriminator optimizes parameters continuously so that whether the image is from fuzziness removing image distribution or clear image distribution can be discriminated more effectively, thegenerator comprises a down-sampling device and an up-sampling device, the down-sampling device carries out convolution on the image and extract semantic information from the image, and the up-sampling device carries out deconvolution on the image by combining the obtained semantic information with structure information of the image. Thus, motion blur of the image can be removed effectively, and the clear image satisfying perception of humans is obtained.

Description

technical field [0001] The invention relates to the technical field of generative confrontation networks, in particular to a method for image motion blur removal based on generative confrontation networks and a generative confrontation network model for motion blur removal. Background technique [0002] Image deblurring technology, that is, deblurring an input image with motion blur noise to generate a clear image after removing blur noise. Image de-blurring technology has a wide range of applications in areas such as unmanned driving, public security investigation, and media processing. For example, in an unmanned driving system, for the captured image of a vehicle with motion blur and noise due to relatively fast motion, the image of the vehicle becomes clearer by applying image de-motion blur technology, thereby Improve the recognition rate of obstacles such as vehicles, thereby effectively improving the safety performance of the unmanned driving system. [0003] Existi...

AI Technical Summary

Problems solved by technology

[0008] The method of solving the blur kernel based on the neural network, after obtaining the blur kernel, still needs to apply the traditional energy equation optimization algorithm to solve the final clear image, which makes the performance of this method be affected by the traditional non-blind deblurring algorithm. limit

In addition, when the fuzzy kernel is solved incorrectly, it will greatly affect the quality of the deblurred image, and the accuracy of the fuzzy kernel solved by the neural network will also greatly restrict the final output of the algorithm.

[0009] Based on the end-to-end neural network method, due to the lack of structural information constraints in most cases during the network training process, the generated clear images are not realistic enough and the visual effect is poor

Images