A Modified Framework for No-Reference Digital Image Quality Assessment

Abstract

In light of the information revolution and rapid technological jumps which caused to increase use the digital images to develop large applications in visual communication systems, the target image is typically degraded from an original perfect quality image is called the reference image. Image quality assessment (IQA) is a measure to assess the quality of an image in understanding or in reference to the original image. May occur degradations in image quality during reproduction, transmission. And also the signal transformed might be exposed to various sorts of distortions which degrade the quality of the image. In this research, use a “reference” called the pseudo reference image (PRI) and introduces PRI based on blind image quality assessment (BIQA) framework. Specific-distortion estimated for PRI based blokiness, PRI based sharpness, and PRI based noisiness after preprocessing of all stages of the model. After extracting features (PSS, LSSs, and LSSn) are using these features to train the classifier (NNK) to identify the general distortion. Through 2-stages to estimate the specific-distortion after preprocessing and then integrating distortion identification. The proposed framework (BPRI) is simplifying by abolishing a score alignment. Can be results obtained (quality score) of images using two datasets LIVE and CSIQ. Comparing the results BPRI proposed framework with subjective image test (DOMS standard).