The use of imaging devices obtain massively large amount of images every day and the Internet makes sharing of these images easier and faster. Digital images are undergoes distortions as it goes through the whole procedure like acquisition, storage, transmission, processing and compression. This makes image quality assessment important in modern systems. According to the availability of original undistorted image, the IQA metric can be classified into three categories, the FullReference (FR), Reduced Reference (RR) and blind/noreference (NR) IQA. Traditional blind image quality assessments predict the quality from a whole distorted image directly. In this paper, multiple pseudo reference images (MPRIs) for NR-IQA is introduced initially by distortion aggravation. For that, the distorted images are subjected to various types of commonly encountered distortions and for each type, five different levels of distortions is added. Later modified local binary patterns(MLBP) features are extracted to describe the similarities between the distorted image and the MPRIs. These similarities metrics are used for estimating the quality of the image using SVM. More similar to a particular pseudo reference image indicates closer to the quality to this PRI. The influence on image content can be reduced by the availability of the created MPRIs. Also the image quality can be inferred more accurately and consistently.

Keywords : Blind/No-Reference Image Quality Estimation (BIQA/NR-IQA), Full-Reference IQA (FR-IQA), Image Quality Estimation (IQA), Natural Scene Images (NSI), Reduced Reference IQA (RR IQA), Screen Content Images (SCI).