Histogram-Equalized Hypercube Adaptive Linear Regression for Image Quality Assessment

Abstract

Image Quality Assessment (IQA) becomes intensely salient in several applications, namely, acquisition of images, watermarking, image compression, image transmission, enhancement of images and so on, due to the extensive use of digital images. In the past decades, considerable advancements have been developed in IQA using Region of Interest (ROI). However, ROI localization is a labour-intensive process that takes multiple passes of sliding-window in search of proper ROI. The efficiency of examination, reduction in the time taken for ROI localization by multiple passes and the quality of the image can be improved by the proposed method, Histogram-Equalized Hypercube Adaptive Linear Regression (HE-HALR) scheme. HE-HALR scheme first performs the pre-processing step for input images. In this step, the features used to describe the quality of images are analysed using Histogram-Equalization-based Contrast Masking (HE-CM) model. The HE-CM model performs ROI localization with the parallelization programming that identifies the contrast masking and luminance value in a parallel manner. With the resultant feature vectors, dimensional reduction is performed using machine learning technique, namely, hypercubical neighbourhood. Finally, IQA is performed with the dimensionality-reduced features using Adaptive Linear Regression.