Multi-Scale Differential Error: A Novel Image Quality Assessment Tool

F. Kanters, L. Florack, B. Platel, and B.t.H. Romeny (The Netherlands)


Image quality, Multi-scale, Error measure


The human visual system is quite capable of judging image quality between images with varying degradation. How ever, most commonly used computer vision methods to calculate a perceptually reasonable error between a refer ence image and some degraded image have problems with varying types of degradation. In this paper we propose a simple, novel method for calculating the error between a reference image and a degraded image that is inspired by the human visual system and is capable of calculating com parable errors for diverse degradations. Experiments show superior performance compared to commonly used, simple mathematical approaches, and comparable results to more complex human visual system based methods, but does not require any parameters or heuristic information.

Important Links:

Go Back