h"p://icv.ims.ut.ee shb@ut.ee Conventional - - PowerPoint PPT Presentation

h"p://icv.ims.ut.ee ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡shb@ut.ee ¡

Conventional Image Enhancement to High Dynamic Range Image Enhancement

• Assoc. Prof. Dr. Gholamreza Anbarjafari

Shahab iCV Research Group

Image Enhancement

Resolution Enhancement Illumination Enhancement Denoising

• In visual perception of the real world,

contrast is determined by the difference in the color and brightness

• f the object with other objects in the

same field of view.

• The human visual system is more

sensitive to contrast than absolute luminance; hence, we can perceive the world similarly regardless of the considerable changes in illumination conditions.

A face image from the CALTECH face database (a), its histogram (b), the equalized face image using GHE (c) and its respective histogram (d).

(b) (a) (c)

(d)

ILLUMINATION

(a) (b)

A face image from the CALTECH face database (a), and the equalized face image using histogram equalization in each R, G, and B channels separately (b).

SINGULAR VALUE DECOMPOSITION

T A A A

A U V = Σ

where UA and VA are orthogonal square matrices known as hanger and aligner respectively, and ΣA matrix contains the sorted singular values on its main diagonal.

ΣA contains the intensity information of the given image

(a) (b) (c) (d) (e) (f) (g) (h)

A grey scale image (a) and the effect of changing the σ1: σ1=0 (b), σ1= σ1+3√σ1 (c), σ1= σ1-3√σ1 (d), σ1= σ1+10√σ1 (e), σ1= σ1-10√σ1 (f), σ1= σ1+0.75σ1 (g), and σ1= σ1-0.75σ1 (h).

SVD BASED EQUALİZATİON: SVE

= Σ

T A A A

A U V

( )

( )

( )

µ

ξ

= =

Σ = Σ

0,var 1

max max

N A

( )

ξ Ξ = Σ

T equalized A A A

A

U V

SVE

(a)

100 200 300 200 400 600 800 1000

(e) (b)

100 200 300 500 1000 1500 2000

(f) (c)

100 200 300 100 200 300 400 500 600 700 800

(g) (d)

100 200 300 200 400 600 800 1000

(h)

A face image from Caltech database (a), introduced low density of the same image (b) and the resultant image of SVE (c) and GHE (d) and their respective smoothed histograms (e)-(h).

WAvelet

Discrete Wavelet Transform Single Tree Complex Wavelet Transform Dual Tree Complex Wavelet Transform 1 Level DWT

DWT

DWT+SVE

LL subband concentrates the illumination information There are two significant parts of the proposed method:

• The first one is the use of SVD. Changing singular

values will directly affect the illumination of the image hence the other information in the image will not be changed.

• The second important aspect of this work is the

application of DWT.

DWT+SVE

Low contrast input satellite image Equalized image using GHE DWT DWT LL LH HL HH HH HL LH LL

Calculate the U, Σ, and V for LL subband image and find the maximum element in Σ. Calculate the U, Σ, and V for LL subband image and find the maximum element in Σ.

Calculate ζ using Eq (4) Calculate the new Σ and reconstruct the new LL image, by using Eq (6). IDWT Equalized satellite image

ζ = max ΣLL ˆ

A

( )

max ΣLLA

( )

ζ

Σ = Σ

= Σ

LL LL

LL LL LL

A

A A LL U V A A A

DWT+SVE

(a) (b) (c) (d) (e) (f)

¡ Original low contrast images from Antarctic Meteorological Research

Centre (a), equalized image by using: GHE (b), LHE (c), SVE (d), BPDHE (e), and proposed technique (f).

DWT+SVE

(a) (b) (c) (d) (e) (f)

¡

Original low contrast image from Satellite imaging Corporation (a), equalized image by using: GHE (b), LHE (c), SVE (d), BPDHE (e), and proposed technique (f).

PUblished Work

1. Demirel, H., & Anbarjafari, G. (2008). Pose invariant face recognition using probability distribution functions in different color channels. Signal Processing Letters, IEEE Signal Processing Letters, IEEE, 15, 537-540. 2. Demirel, H., Anbarjafari, G., & Jahromi, M. N. S. (2008, October). Image equalization based on singular value decomposition. In In Computer Computer and and Information Information Sciences Sciences, 2008. ISCIS'08. 23rd International Symposium on (pp. 1-5). IEEE IEEE. 3. Demirel, H., Ozcinar, C., & Anbarjafari, G. (2010). Satellite image contrast enhancement using discrete wavelet transform and singular value decomposition. Geoscience Geoscience and and Remote Remote Sensing Sensing Letters, IEEE Letters, IEEE, 7(2), 333-337. 4. Anbarjafari, G., Jafari, A., Jahromi, M. N. S., Ozcinar, C., & Demirel, H. (2015). Image illumination enhancement with an objective no- reference measure of illumination assessment based on Gaussian distribution mapping. Engineering Engineering Science Science and and Technology Technology, an International Journal, 18(4), 696-703.

PUblished Work

5. Ozcinar, C., Demirel, H., & Anbarjafari, G. (2011). Image Equalization Using Singular Value Decomposition and Discrete Wavelet

• Transform. Discrete

Discrete Wavelet Wavelet Transforms: Transforms: Theory Theory and and Applications Applications, 87-94. 6. Anbarjafari, G., Izadpanahi, S., & Demirel, H. (2015). Video resolution enhancement by using discrete and stationary wavelet transforms with illumination compensation. Signal, Signal, Image Image and and Video Video Processing Processing, 9(1), 87-92. 7. Demirel, H., Anbarjafari, G., Ozcinar, C., & Izadpanahi, S. (2011, September). Video resolution enhancement by using complex wavelet transform. In Image Image Processing Processing (ICIP), 2011 18th IEEE International Conference on (pp. 2093-2096). IEEE IEEE.

HDR

• High Dynamic Range Imaging
• 10-12-14-16-… bits
• Displays are conventional 8-10 bits
• Standards?

HDR

• Collaborative work with Telecom

ParisTech for ICIP2016

• Adaptive HDR display
• Reduction of flickers

HDR

• Demo 1
• Demo 2

HDR

Thank You