A Zoom Image Generation Method Combining Block Matching and Neural Networks

Abstract

The invention discloses a zoom image generation method that combines block matching and neural networks. At the same time, the telephoto lens and the short focus lens of the bifocal camera are used to shoot the same shooting scene or object under the same optical axis to obtain the telephoto image and the short focus image; the Unet neural network structure is established, and the Unet neural network structure includes convolution In the partial deconvolution part, both the telephoto image and the short-focus image are input to the convolution part for image feature extraction, thereby obtaining the high-frequency detail feature structure of the telephoto image and the short-focus image; a block matching algorithm is used to combine the telephoto image and the short-focus image. The high-frequency detail feature structure of the focal image is matched and fused to obtain a cascade image; the cascade image is then input into the deconvolution part of the Unet neural network structure for reconstruction to obtain a zoom image. Based on the requirement of continuous zooming of digital images, the present invention realizes continuous zooming of digital images at any magnification through the structure of image cropping and image deconvolution.

Description

technical field [0001] The invention belongs to a digital image zoom algorithm in the technical field of digital imaging, and in particular relates to a zoom image generation method combined with block matching and neural network. [0002] technical background [0003] It is well known that high-resolution images can provide more details than their low-resolution counterparts. These details should be crucial in all fields such as remote sensing, medical diagnosis, intelligent monitoring, etc. Due to the limitations of optical zoom, digital zoom has been widely used in many imaging devices. Digital zoom digitally enlarges the image without changing the focal length of the lens, thus resulting in a loss of image quality: however, image processing algorithms such as image interpolation used in digital zoom systems, in addition to jaggies and blurring artifacts Does not produce high-quality pictures. To solve this problem, many improved algorithms have been proposed in the pas...

AI Technical Summary

Problems solved by technology

Although it has fast processing time at low computational complexity, the step-by-step approach cannot guarantee accurate estimation, especially in the presence of noise

Some literatures propose to use the method of optical flow matching to fuse images of different focal lengths, but the quality of the output image is not high due to the expensive time cost and the problem that the optical flow matching cannot be fully registered.

Images