Perceptual quality evaluation for image defocus deblurring

摘要: Abstract Blur is one of the most common distortion types in image acquisition. Image deblurring has been widely studied as an effective technique to improve quality blurred images. However, little work done perceptual evaluation algorithms and deblurred In this paper, we conduct both subjective objective studies defocus deblurring. A database (DDID) first built using state-of-the-art algorithms, test carried out collect human ratings Then performances are evaluated based on scores. With observation that existing metrics limited predicting images, a enhancement module proposed Gray Level Co-occurrence Matrix (GLCM), which mainly used measure loss texture naturalness caused by Experimental results DDID demonstrate effectiveness method.