Artifact reduction in JPEG2000 compressed images at low bit-rate using mathematical morphology filtering

L. Bennacer B. Bouledjfane A. Nait-Ali 1
1 SYNAPSE
LISSI - Laboratoire Images, Signaux et Systèmes Intelligents
Abstract : JPEG2000 is known as an efficient standard to encode images. However, at very low bit-rates, artifacts or distortions can be observed in decoded images. In order to improve the visual quality of decoded images and make them perceptually acceptable, we propose in this work a new preprocessing scheme. This scheme consists in preprocessing the image to be encoded using a nonlinear filtering, considered as a prior phase to JPEG 2000 compression. More specifically, the input image is decomposed into low- and high-frequency sub-images using morphological filtering. Afterward, each sub-image is compressed using JPEG2000, by assigning different bit-rates to each sub-image. To evaluate the quality of the reconstructed image, two different metrics have been used, namely (a) peak signal to noise ratio, to evaluate the visual quality of the low-frequency sub-image, and (b) structural similarity index measure, to evaluate the visual quality of the high-frequency sub-image. Based on the reconstructed images, experimental results show that, at low bit-rates, the proposed scheme provides better visual quality compared to a direct use of JPEG2000 (excluding any preprocessing).
Liste complète des métadonnées

https://hal.archives-ouvertes.fr/hal-01568420
Contributor : Lab Lissi <>
Submitted on : Tuesday, July 25, 2017 - 11:43:10 AM
Last modification on : Thursday, February 28, 2019 - 11:07:14 AM

Identifiers

  • HAL Id : hal-01568420, version 1

Collections

Citation

L. Bennacer, B. Bouledjfane, A. Nait-Ali. Artifact reduction in JPEG2000 compressed images at low bit-rate using mathematical morphology filtering. Signal, Image and Video Processing, Springer, 2014, 8 (4), pp.677-686. 〈hal-01568420〉

Share

Metrics

Record views

72