Image enhancement method, model training method and equipment

Abstract

The embodiment of the invention discloses an image enhancement method, a model training method and equipment, which can be applied to the field of image processing in the field of artificial intelligence, and the method comprises the following steps: extracting features from an input image through a first neural network layer to obtain a first feature, performing pixel classification and image classification on the first feature through a second neural network layer and a third neural network layer to generate first classification information and second classification information, and obtaining a target lookup table based on the first classification information, the second classification information and a spatial perception three-dimensional lookup table (3D LUT); building the spatial perception three-dimensional lookup table according to each image category and each pixel category; and finally, obtaining an enhanced image according to the input image and the target lookup table. Compared with a traditional three-dimensional lookup table, the method has the advantages that the processing capacity is improved, and the problem that inaccurate results (such as local wrong colors and artifacts) are easily generated due to less information in an enhancement method based on the traditional three-dimensional lookup table is solved.

Description

technical field [0001] The present application relates to the field of image processing, in particular to an image enhancement method, model training method and equipment. Background technique [0002] Image enhancement is a method to improve the visual effect of an image. Generally speaking, the main purpose of image enhancement is to improve the visual effect of the image and improve the clarity of the image. Resolution reconstruction etc. Traditional image enhancement methods are mainly divided into methods based on the space domain and frequency domain. The method based on the space domain is to directly process the image; the method based on the frequency domain is to correct the transformation coefficient of the image in a certain transformation domain of the image, and then Inversely transform to the original space domain to obtain an enhanced image. [0003] With the development of deep learning technology, the enhancement effect of image enhancement methods based ...

AI Technical Summary

Problems solved by technology

[0005] The embodiment of this application provides an image enhancement method, a model training method and equipment, which are used to construct a The space-aware 3D lookup table obtains the target lookup table, and based on the target lookup table, the input image is enhanced, which improves the processing ability compared with the traditional 3D lookup table, and solves the problem of lack of information in the enhancement method based on the traditional 3D lookup table (only pixel value) which is prone to the problem of inaccurate results (eg, locally produced wrong colors, artifacts, etc.)

Images