2006
DOI: 10.1016/j.image.2006.08.003
|View full text |Cite
|
Sign up to set email alerts
|

Low-complexity compression of multispectral images based on classified transform coding

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
2
1

Citation Types

1
10
0

Year Published

2010
2010
2024
2024

Publication Types

Select...
6
1

Relationship

0
7

Authors

Journals

citations
Cited by 15 publications
(11 citation statements)
references
References 14 publications
1
10
0
Order By: Relevance
“…The Karhunen-Loève transform has proven to be especially suitable to decorrelate spectral information [17]- [19]. Meanwhile, efficient approaches to exploit the intrinsic nature of remote sensing images have been proposed, including pixel classification [20] and models of anomalous pixels [21]. Again, JPEG2000 is widely known in the community due to the provision of advanced features such as multi-component transforms and effective interactive transmission protocols.…”
mentioning
confidence: 99%
See 1 more Smart Citation
“…The Karhunen-Loève transform has proven to be especially suitable to decorrelate spectral information [17]- [19]. Meanwhile, efficient approaches to exploit the intrinsic nature of remote sensing images have been proposed, including pixel classification [20] and models of anomalous pixels [21]. Again, JPEG2000 is widely known in the community due to the provision of advanced features such as multi-component transforms and effective interactive transmission protocols.…”
mentioning
confidence: 99%
“…Therefore, coding systems should be devised while keeping in mind the device in which they will be executed [29], [30]. There are many works concerned with the computational complexity of techniques and algorithms [31], [32] deployed to code 3D images [11], [14], [19], [20], [22], [33]- [36].…”
mentioning
confidence: 99%
“…On the other hand, static thresholds guarantee stable quality with still good computational savings. The most remarkable structures on the candidate list are the one with static thresholding with the following configuration: [(32, 7, 2), (4,16,6), (2,12,4), (1,8,0)] and the dynamic one based on AE and a cluster size of four (AE size 4). The former achieves almost the same quality as the full KLT, while keeping cost lower than the previous approaches, and still with a decent scalability.…”
Section: Resultsmentioning
confidence: 99%
“…This strategy is compatible with our proposal and will be discussed later. Another strategy to reduce training computational costs is to train once the KLT for a training set of images for all images [6], although with afterward, employ the already trained transform for all images, although with this strategy, the transform is not specifically trained for each image, and its coding performance is worse.…”
Section: Introductionmentioning
confidence: 99%
“…Due to this similarity, the subspace dimensionality of a cluster can be lower than that of the whole image. In this regard, Kaarana et al [4] and Cagnazzo et al [5] classified pixels into regions according to spectral similarity, and compressed the regional-spectrum individually. Chang [6] divided multispectral satellite image into eigenregions.…”
Section: Introductionmentioning
confidence: 99%