Image deblurring method and device, and equipment

Abstract

The invention discloses an image processing method and device, a medium and equipment, relates to the field of artificial intelligence, and is used for removing a moving target in an image, and the method specifically comprises the following steps: extracting original image information of three channels from a blurred image; performing Laplace transformation and mean square error removal calculation on the original image information of the three channels to obtain intermediate image information corresponding to the three channels; averaging the intermediate image information respectively corresponding to the three channels to obtain the ambiguity of the blurred image; and selecting a corresponding deep learning network model to perform deblurring operation on the blurred image according to the blurring degree of the blurred image.

Description

technical field [0001] The present invention relates to the technical field of image processing, in particular to an image deblurring method, device and equipment. Background technique [0002] In recent years, dual-lens or even multi-lens smart cameras have been widely used, such as smart dual-camera mobile phones, unmanned driving, drones, etc. However, the problem of out-of-focus blur cannot be avoided in many scenarios, such as equipment Shake, misfocus, or fast-moving objects during shooting can produce image blur. This will not only affect the quality of the captured pictures, but also have a serious impact on the subsequent processing of the pictures. There are many traditional deblurring algorithms, which can be divided into blind deconvolution algorithms and non-blind deconvolution algorithms according to whether there is a priori model. The former has certain a priori assumptions about the convolution kernel or the image, and the latter assumes that the blur kern...

AI Technical Summary

Problems solved by technology

The former has certain a priori assumptions about the convolution kernel or the image, and the latter assumes that the blur kernel function is known. These two traditional algorithms have very difficult problems to solve. In the case of unknown blur type and deep blur It is often difficult to achieve satisfactory results

[0003] The current popular image deblurring algorithms often require a large convolutional neural network. Although the convolutional neural network can handle various types of distortion or blurred images of different degrees, choosing a network model with high complexity will bring large computational overhead

In addition, since these algorithms use the same neural network structure to process all images, it is still difficult to achieve satisfactory results after some images are processed.

Images