No-reference type super-resolution image quality evaluation method based on stacking

Abstract

The invention discloses a no-reference type super-resolution image quality evaluation method based on stacking, and the method specifically comprises the following steps: firstly extracting the depthfeatures of a super-resolution image through an existing trained VGGnet model, and quantifying the degradation of the super-resolution image; then, using a stepping regression algorithm containing anSVR algorithm and a k-NN algorithm as a first-layer regression model; and constructing a mapping model from the depth features extracted from the VGGnet model to the prediction quality score, and obtaining a second-layer regression model by adopting a linear regression algorithm, thereby forming a staging regression model to realize non-reference evaluation of the quality of the super-resolution image. According to the method, the complementarity advantages of two different basic regression devices of SVR and k-NN are utilized, linear regression is used as a meta-regression device, the prediction accuracy is improved, and the super-resolution image quality score closer to human eye subjective evaluation can be obtained.

Description

technical field [0001] The invention belongs to the technical field of image processing and analysis methods, and relates to a stacking-based non-reference super-resolution image quality evaluation method. Background technique [0002] Single-frame image super-resolution reconstruction is a technique that utilizes information from one or more low-resolution input images to generate a high-resolution image with finer details. This technology is widely used in image processing, computer vision and other fields. With the emergence of a large number of super-resolution image reconstruction algorithms, how to evaluate image super-resolution reconstruction algorithms has become a key research issue. There is no doubt that human vision is the ultimate receptor for evaluating images, so subjective quality assessment is the most direct and effective method to reflect the quality of super-resolution images. However, the procedure of the subjective quality assessment method is time-c...

AI Technical Summary

Problems solved by technology

[0005] The purpose of the present invention is to provide a super-resolution image quality evaluation method based on stacking without reference, which solves the problem that the super-resolution image quality evaluation method existing in the prior art is difficult to accurately and effectively evaluate the super-resolution image quality

Images