Full-reference and no-reference image quality assessment methods with unified structure

Abstract

The invention discloses a full-reference and no-reference image quality evaluation method with a unified structure, and relates to image processing. The method includes; Carrying out image segmentation and filtering decomposition processing; And designing a feature extraction network and a regression network. The distortion image quality can be evaluated, and the method can be applied to full-reference and no-reference image quality evaluation tasks. A convolutional neural network is adopted to learn the mapping from the image features to the main pipe quality score, thereby realizing an imagequality evaluation task. The method is established on the basis of deep learning machine learning set statistics, and has the characteristics that the structure of different image quality evaluationtasks is unchanged, the practicability and reliability are high, the portability is high, the subjective quality score conforming to human eye aesthetic appreciation can be predicted, and the labor cost of human eye evaluation is saved.

Description

technical field [0001] The present invention relates to image processing, in particular to a full-reference and no-reference image quality evaluation method based on a unified structure of deep learning and statistics. Background technique [0002] Image quality evaluation is an important basic work in image processing, and it has applications in many other image processing tasks, such as image enhancement and super-resolution reconstruction, so it has attracted the attention of researchers. The image quality evaluation task is divided into: no reference image quality evaluation (distorted image does not exist reference image), semi-reference image quality evaluation (reference image exists in part of the distorted image), full reference image quality evaluation (all distorted images are reference image exists). The difficulty of achieving these three tasks decreases in turn. Literature ([1] Zhang L, Zhang L, Mou X, etal. FSIM: A feature similarity index for image quality ...

AI Technical Summary

Problems solved by technology

The method of image quality evaluation based on artificial features is difficult to achieve good generalization ability

The neural network has a strong nonlinear fitting ability, so it can learn features with strong generalization ability, Gao F([4]Gao F,Wang Y,Li P,Tan M,Yu J,Zhu Y.Deepsim:Deep similarity for image quality assessment[J].Neurocomputing,2017,257:104-114) and Bosse S([5]Bosse S,Maniry D,Müller K R,et al.Deep neural networks for no-reference and full-reference image quality assessment[ J]. IEEE Transactions on Image Processing, 2018, 27(1): 206-219) proposes an image quality evaluation method based on convolutional neural network. The predicted score has a high consistency with the actual subjective score, but these methods ignore the Application of image domain knowledge

In addition, the network structure of deep learning for no-reference image quality assessment tasks and full-reference image quality assessment tasks is different

Images