Image quality measurement using integer wavelet transformations

Abstract

We present a novel and straightforward method of graphical and scalar image quality measurement utilising integer wavelet transformations. The measure can represent separately the components of image distortion such as noise and blur through relative energy in the wavelet transform subbands. The measure can perform a similar function to the Hosaka plot whilst not requiring segmentation and threshold parameters. Through weighted summation, the graphical measure can be degenerated into a scalar quality measure. The scalar measure is compared with other image quality measures from recent literature and found to present consistently high correlation with subjective image quality assessment using NIIRS.