Non-reference quality evaluation algorithm combining multiple edge detection operators

Abstract

The invention relates to a non-reference quality evaluation algorithm combining multiple edge detection operators. A combination entropy and a chain-type rule are used to define a conditional entropy.The algorithm comprises the following steps of selecting an image used for training and testing; calculating the gradient of the image, a relative gradient, and a Gaussian Laplace operator (LoG), wherein extracted image characteristics include five-dimensional characteristics of the standard deviation of a relative gradient direction RO, the conditional entropies H(GM|L) and H(L|GM) between the gradient and LoG, and the conditional entropies H (GM|L)-H (RM|L) and H(L|GM)-H(L|RM)) between the relative gradient and the LoG; according to multi-scale performance in a human eye visual system characteristic, using a downsampling method, extracting the above five-dimensional characteristics of the reduced image, and finally acquiring a ten-dimensional characteristic vector; and using AdaBoost neural network to carry out regression so as to predict an image quality score.

Description

Technical field: [0001] The invention relates to the field of reference-free objective quality evaluation of 2D digital images. Background technique: [0002] With the rapid development of digital imaging technology and the Internet, images are used more and more frequently in people's daily life. However, images will inevitably be distorted during the process of collection, compression, transmission, storage, etc., resulting in a certain degradation in the final image, which cannot meet people's needs. Designing an effective image quality assessment method (ImageQuality Assessment, IQA) plays an important role in image processing such as image compression, image deblurring, and image enhancement. Image quality evaluation methods can be divided into two categories: subjective evaluation and objective evaluation. The former is to score the quality of the picture by several observers, and then obtain the average score through statistical methods, which is called MOS (Mean Op...

AI Technical Summary

Problems solved by technology

At present, algorithms based on edge information have achieved good results. Most of these methods use a single edge detection algorithm, which has certain pertinence, but it is difficult to reflect all the texture details of the image.

Images