To improve evaluation performance of the non-reference image quality evaluation method

an image quality evaluation algorithm applying image re-fuzzy theory is proposed.The image to be tested is subjected for re-fuzzy process

extract the local and global features of the image by using difference between the two images before and after blurring

co-building multi-dimensional feature vectors

and the support vector regression（SVR）method is used to realize the non-reference quality evaluation of image.Test experiments in public image database show that the image quality evaluation method has higher accuracy and better generalization ability

and has better consistency with human subjective evaluation.