Napredna pretraga

Pregled bibliografske jedinice broj: 91508

An Algorithm for a Fast Estimation of the Wavelet Subbands Statistics


Rimac-Drlje, Snježana; Zovko-Cihlar, Branka; Grgić, Sonja
An Algorithm for a Fast Estimation of the Wavelet Subbands Statistics // Proceedings of the 3rd EURASIP Conference focused on Digital Signal Processing for Multimedia Communications and Services, ECMCS'2001 / Kalman Fazekas (ur.).
Budapest, Hungary: Scientific Association of Infocommunications - HTE, 2001. str. 284-287 (predavanje, međunarodna recenzija, cjeloviti rad (in extenso), znanstveni)


Naslov
An Algorithm for a Fast Estimation of the Wavelet Subbands Statistics

Autori
Rimac-Drlje, Snježana ; Zovko-Cihlar, Branka ; Grgić, Sonja

Vrsta, podvrsta i kategorija rada
Radovi u zbornicima skupova, cjeloviti rad (in extenso), znanstveni

Izvornik
Proceedings of the 3rd EURASIP Conference focused on Digital Signal Processing for Multimedia Communications and Services, ECMCS'2001 / Kalman Fazekas - Budapest, Hungary : Scientific Association of Infocommunications - HTE, 2001, 284-287

Skup
3rd EURASIP Conference focused on Digital Signal Processing for Multimedia Communications and Services, ECMCS'2001

Mjesto i datum
Budimpešta, Mađarska, 11-13.09.2001

Vrsta sudjelovanja
Predavanje

Vrsta recenzije
Međunarodna recenzija

Ključne riječi
Wavelet Transform; Image Compression; Probability Density Function; Quantization Noise

Sažetak
Discrete wavelet transform - DWT is successfuly used for image compression in different coders. The knowledge of DWT coefficients statistics can improve behavior of the coding techniques. The paper proposes a fast algorithm for estimation of the wavelet coefficients probability density function - PDF. Wavelet subband PDF is approximated well by generalized Gaussian function, but parameter k of that function widely differs for different subbands, and it also differs for different images. The parameter k determines the shape of the PDF and it influences compression ratio and quantization noise power in the image compression procedure. The proposed algorithm is based on the quantization noise power dependance on the parameter k. We show the usage of the estimated PDF for calculation of the mass center of quantization intervals, which we used as reconstruction levels. So, we obtained an improvement in PSNR (Peak Signal-to-Noise Ratio) of reconstructed images between 0.25 and 0.4 dB.

Izvorni jezik
Engleski

Znanstvena područja
Elektrotehnika



POVEZANOST RADA


Projekt / tema
036015
036041

Ustanove
Fakultet elektrotehnike i računarstva, Zagreb