# A Universal Image Quality Index Author Zhou Wang

• Slides: 15

A Universal Image Quality Index Author : Zhou Wang, Student Member, IEEE, and Alan C. Bovik, Fellow, IEEE Source : IEEE SIGNAL PROCESSING LETTERS, VOL. 9, NO. 3, MARCH 2002 Advisor : Chin-Chen Chang Speaker : Yi-Chang Liu Date : 2004/4/21

Outline • • Introduction Definition of The New Quality Index Application to Images Conclusion and Discussion 2020/12/4 2

Introduction • Two classes of objective quality or distortion approaches • Mathematically：Mean Squared Error (MSE)、 Peak Signal to Noise Ratio (PSNR) • Human Visual System (HVS) 2020/12/4 3

Introduction (con. ) • Why mathematically defined measures? • • Easy to calculate Usually have low computational complexity Independent of viewing Individual observers 2020/12/4 4

Introduction (con. ) • Universal • The images being tested • The viewing conditions • The individual observers 2020/12/4 5

Definition of The New Quality Index (The original image signals) • (The change image signals) The proposed quality index 2020/12/4 6

Definition of The New Quality Index (con. ) • Where • The dynamic range of Q is [-1, 1] • • 2020/12/4 7

Definition of The New Quality Index (con. ) • (The original image Ex. signals)20 30 40 2020/12/4 (The change image signals) 20 30 40 50 60 70 80 90 10 8

Definition of The New Quality Index (con. ) • Rewrite the definition of Q [-1, 1] Loss of correlation 2020/12/4 [0, 1] Luminance distortion contrast distortion 9

Definition of The New Quality Index (con. ) • Loss of correlation • The best value 1 is obtained when for all i=1, 2, …, N , where a and b are constants and a > 0 • Luminance distortion • It equals 1 if and only if • contrast distortion • The best value 1 is achieved if and only if 2020/12/4 10

Application to Images If there a total of M steps , then the overall quality index is given by： 2020/12/4 11

Application to Images (con. ) Original 2020/12/4 Additive Gaussian Noise Impulsive Salt-Pepper Noise Multiplicative Speckle Noise 12

Application to Images (con. ) Mean Shift 2020/12/4 Blurring Contrast Stretching JPEG Compression 13

Application to Images (con. ) Distortion Type Mean Subjective Rank MSE Q Mean Shift 1. 59 225 0. 9894 Contrast Stretching 1. 64 225 0. 9372 Impulsive Salt-Pepper Noise 3. 32 225 0. 6494 Multiplicative Speckle Noise 4. 18 225 0. 4408 Additive Gaussian Noise 4. 27 225 0. 3891 Blurring 6. 32 225 0. 3461 JPEG Compression 6. 68 215 0. 2876 2020/12/4 14

Conclusion and Discussion • • • Outperforms the MSE Simple Sensitive to the energy of errors 2020/12/4 15