No-Reference Image Quality Assessment Using the Statistics of Global and Local Image Features

Methods of image quality assessment are widely used for ranking computer vision algorithms or controlling the perceptual video and streaming applications. The ever-increasing number digital images has encouraged research in this field at an accelerated pace recent decades. After appearance convolutional neural networks, many researchers have paid attention to different deep architectures devise no-reference algorithms. However, systems still rely on handcrafted features ensure interpretability restrict consumption resources. In study, our efforts focused creating a quality-aware feature vector containing information about both global local features. Specifically, results visual physiology indicate that human system first quickly automatically creates perception before gradually focusing certain areas judge image. broad spectrum statistics extracted from is utilized represent aspects various points view. experimental demonstrate method’s predicted ratings relate strongly with subjective ratings. particular, introduced algorithm was compared 16 other well-known advanced methods outperformed them by large margin 9 accepted benchmark datasets literature: CLIVE, KonIQ-10k, SPAQ, BIQ2021, TID2008, TID2013, MDID, KADID-10k, GFIQA-20k, which considered de facto standards generally assessment.