Content-based image quality assessment
Peer reviewed, Journal article
Accepted version
Åpne
Permanent lenke
https://hdl.handle.net/11250/2677556Utgivelsesdato
2020Metadata
Vis full innførselSamlinger
Originalversjon
International journal of imaging and robotics. 2020, 20 (3), 39-71.Sammendrag
In this work we investigate the influence of image content on image quality assessment. We describe image content through a set of features, such as busyness, colorfulness, lightness, sharpness and entropy, investigating the influence of each feature separately and
their combination on the performance of image quality metrics. Based on the found dependencies we developed prediction models based on SVM and Random Forest machine learning techniques, which predict correctly and wrongly estimated images (compared to observer score) by an image quality metric.