A semi-reference image quality assessment method based on spectral residual

Abstract

The present invention belongs to the image assessment technology field, and provides a reduced-reference image quality assessment method based on spectrum residual errors. The method comprises the following steps: calculating the RSR(Id) values and the RSR(Ir) values of an image Id to be assessed and a reference image Ir; calculating the RSRM values of the image to be assessed and the reference image; assessing the image quality according to the sizes of the RSRM values, wherein the calculation process of the RSR values of the image to be assessed and the reference value comprises: calculationof the gradient size G(I) of an image I, calculation of wavelet coefficient {DWT(I), DWT(G(I))} of the image I and the G(I), calculation of the spectrum residual errors SR of each component in the wavelet coefficient {DWT(I), DWT(G(I))}, and calculation of the significant values SM of each component in the wavelet coefficient {DWT(I), DWT(G(I))}; performing encoding through fractal dimension, andobtaining the residual spectrum regularity value RSR of the significant values SM of each component. The image assessment method has good performance and can be widely applied to the image processingfield.

Description

technical field [0001] The invention relates to a semi-reference image quality evaluation method, in particular to a semi-reference image quality evaluation method based on spectral residual. Background technique [0002] Image quality assessment (IQA) is a fundamental problem in image processing and is key to many applications, such as perceptually optimal encoder design, image communication, and image restoration, etc. According to the availability of raw images, IQA can be divided into three categories: full-reference IQA (FR-IQA), no-reference IQA (NR-IQA) and semi-reference IQA (RR-IQA). FR-IQA requires the entire original image, NR-IQA does not require any information from the original image, and RR-IQA requires partial information from the original image. For RR-IQA, the core problem of measurement is feature extraction. The features of an image can be local or global. Global features describe the visual characteristics of the entire image. For example, Wang, Z., ...

AI Technical Summary

Problems solved by technology

One problem is that some of the features used by the RR-IQA metric are extracted directly from the image and do not take into account that not all distortions are distinguishable in the human visual system, e.g. SRRM methods

Another problem is that some image features used by the RR-IQA metric are not robust to different distortion types

They perform well for images that share the same distortion type, but they don't perform well when multiple distortion types are involved, such as WNISM

Images