How many operations are required to compute the DTW similarity measure?

The authors have experimentally found that the computation of the DTW similarity measure requires 10 times more operations than the computation of the Euclidean distance.

Why is the aim of this article not to get the elongation results?

As stated previously, the aim of this article is not to get the best elongation results, but to be able to compute correct elongations with a low computational cost.

What is the problem with the standard distance used to compare two histograms?

this distance suffers from the problem of the shuffling invariance property [45] which is not desirable when comparing two histograms of ordinal type measurements.

How can the authors detect complex patterns without the noise induced by the details?

By analysing first the image content at a coarse resolution and then gradually increasing this resolution, it is possible to detect complex patterns while avoiding the semantic noise induced by the details, as proposed, e.g., in [38].

What is the way to deal with the variability in the image?

To deal with this variability, two solutions can be considered: (1) to use size/scale invariant features, or (2) to process the image at different spatial resolutions.

What is the histogram of The authorassociated to a subset of X ?

The histogram of The authorassociated to a subset X ⊆ E is the function HI,X : V → N which associates to each value v ∈ V the numberHI(v) = |I−1({v}) ∩ X|, i.e., the histogram of the restriction of The authorto X.

What are the main reasons why the proposed method cannot be compared?

the methods to which the proposed one could be compared, and in particular those considered here (namely, [49, 7, 50] for the extraction of basic objects, and [6] for the extraction of complex objects) can not be directly compared on the datasets described above, especially due to the unavailability of the corresponding softwares and the difficulty to correctly re-implement them by only using the publication details.

Why is it important to develop tools to extract complex patterns from VHSR images?

In this context, and due to the actual importance to analyse VHSR images [4] in addition to lower spatial resolution ones, it is then relevant to develop tools adapted to the extraction of complex patterns from such data, and in particular (low-level) segmentation ones.

Open AccessJournal Article10.1016/J.PATCOG.2011.07.017

Extraction of complex patterns from multiresolution remote sensing images: A hierarchical top-down methodology

Camille Kurtz, +3 more

- 01 Feb 2012

- Pattern Recognition

- Vol. 45, Iss: 2, pp 685-706

TL;DR: A hierarchical approach is proposed to progressively extract segments of interest from the lowest to the highest resolution data, and then finally determine urban patterns from VHSR images, inspired by the principle of photo-interpretation.

About: This article is published in Pattern Recognition. The article was published on 01 Feb 2012. and is currently open access. The article focuses on the topics: Segmentation.

AI Agents for this Paper

Find similar papers on Google Scholar, PubMed and Arxiv
Write a critical review of this paper
Analyze citations of this paper to find unaddressed research gaps

Most frequently asked questions

1. What is the way to extract the relevant nodes?

In order to extract the most relevant nodes which characterise this cut, it is necessary to first separate linear elements from those which correspond to the areas “bounded” by these linear elements.

2. What is the DTW similarity measure used to compare two histograms?

The computational complexity required to compare two histograms HI1 , HI2 : V → N when using the Euclidean distance is in Θ(card(V)) while the complexity is in Θ(card(V) × c) (where c measures the tolerance of distortions on the radiometric axis) when using the constraint DTW similarity measure.

3. What is the way to learn the user’s behaviour from a segmentation example?

By opposition to other strategies devoted to automatically extract cuts from partition hierarchies [7, 42], with the risk of generating non-relevant results, the authors propose to learn the user’s behaviour from a segmentation example.

4. How many operations and time to process a HSR image?

For instance, a HSR image which contains 16 times more pixels than a MSR one, requires 16 times more operations and time to be processed than a MSR one.

Figure 6: Illustration of the proposed example-based segmentation approach: learning of the cut example (see Section 4.1.2)

Table 5: Results evaluation of the Elsau, Illkirch, Hautepierre datasets. The symbol (•) means that the thematic class is not available in the current dataset.

Figure 14: Segmentation results of the Elsau dataset (6 576 m × 2 793 m). The colours of the clusters were chosen to correspond to those defined for the groundtruth maps. (a) MSR segmentation, districts level. (b) HSR segmentation, blocks level. (c) VHSR segmentation, objects level.

Figure 9: Multiresolution clustering approach (see Section 4.2.2).

Table 3: Thematic classes associated to the ground-truth maps. The colours of the classes refer to those of the ground-truth maps of the Elsau dataset (Figure 12).

Table 6: Runtime and memory usage for the segmentation of the images contained in the studied datasets.

Citations

•Journal Article•10.3390/RS11121443

Unmanned Aerial Vehicle for Remote Sensing Applications—A Review

Huang Yao, +2 more

- 18 Jun 2019

- Remote Sensing

TL;DR: This paper performs a critical review on RS tasks that involve UAV data and their derived products as their main sources including raw perspective images, digital surface models, and orthophotos and focuses on solutions that address the “new” aspects of the U drone data including ultra-high resolution; availability of coherent geometric and spectral data; and capability of simultaneously using multi-sensor data for fusion.

...read moreread less

540

Journal Article•10.1016/J.RSE.2012.09.014

Carotenoid content estimation in a heterogeneous conifer forest using narrow-band indices and PROSPECT + DART simulations

Rocío Hernández-Clemente, +2 more

- 01 Dec 2012

- Remote Sensing of Environment

TL;DR: In this paper, the use of narrow-band indices formulated in the visible spectral region at leaf and canopy levels to estimate carotenoid content was explored in a pine forest affected by decline processes.

...read moreread less

152

•Journal Article•10.3390/COMPUTERS8040077

Review on Techniques for Plant Leaf Classification and Recognition

Muhammad Azfar Firdaus Azlah, +4 more

- 21 Oct 2019

- The first computers

TL;DR: This paper aims to review and analyze the implementation and performance of various methodologies on plant classification, finding that each technique has its advantages and limitations in leaf pattern recognition.

...read moreread less

149

Journal Article•10.1080/01431161.2018.1513666

Very high resolution remote sensing image classification with SEEDS-CNN and scale effect analysis for superpixel CNN classification

Xianwei Lv, +3 more

- 17 Jan 2019

- International Journal of Remote Sensing

TL;DR: The superpixel extraction via SEEDS method was found to be the optimal superpixel segmentation approach for CNN classification, and the scale effect on CNN classification accuracy was investigated by comparing the four super pixel segmentation methods.

...read moreread less

125

Journal Article•10.1109/TGRS.2014.2367129

Marker-Controlled Watershed-Based Segmentation of Multiresolution Remote Sensing Images

Raffaele Gaetano, +4 more

- 01 Jun 2015

- IEEE Transactions on Geoscience and Remo...

TL;DR: Numerical results on object layer extraction and simple classification tasks prove the proposed techniques to provide accurate segmentation maps, which preserve fine details and, contrary to state-of-the-art products, can single out objects equally well at very different scales.

...read moreread less

118

...

Expand

References

•Journal Article•10.1109/34.192463

A theory for multiresolution signal decomposition: the wavelet representation

Stéphane Mallat

- 01 Jul 1989

- IEEE Transactions on Pattern Analysis an...

TL;DR: In this paper, it is shown that the difference of information between the approximation of a signal at the resolutions 2/sup j+1/ and 2 /sup j/ (where j is an integer) can be extracted by decomposing this signal on a wavelet orthonormal basis of L/sup 2/(R/sup n/), the vector space of measurable, square-integrable n-dimensional functions.

...read moreread less

22.3K

•Journal Article•10.1109/34.868688

Normalized cuts and image segmentation

Jianbo Shi, +1 more

- 01 Aug 2000

- IEEE Transactions on Pattern Analysis an...

TL;DR: This work treats image segmentation as a graph partitioning problem and proposes a novel global criterion, the normalized cut, for segmenting the graph, which measures both the total dissimilarity between the different groups as well as the total similarity within the groups.

...read moreread less

15.6K

•Proceedings Article•10.1109/CVPR.1997.609407

Normalized cuts and image segmentation

Jianbo Shi, +1 more

- 17 Jun 1997

...read moreread less

14.7K

Journal Article•10.1016/0034-4257(91)90048-B

A review of assessing the accuracy of classifications of remotely sensed data

Russell G. Congalton

- 01 Jul 1991

- Remote Sensing of Environment

TL;DR: This paper reviews the necessary considerations and available techniques for assessing the accuracy of remotely sensed data including the classification system, the sampling scheme, the sample size, spatial autocorrelation, and the assessment techniques.

...read moreread less

7.7K

Journal Article•10.1109/TASSP.1978.1163055

Dynamic programming algorithm optimization for spoken word recognition

H. Sakoe, +1 more

- 01 Feb 1978

- IEEE Transactions on Acoustics, Speech, ...

TL;DR: This paper reports on an optimum dynamic progxamming (DP) based time-normalization algorithm for spoken word recognition, in which the warping function slope is restricted so as to improve discrimination between words in different categories.

...read moreread less

6.7K

...

Expand

Extraction of complex patterns from multiresolution remote sensing images: A hierarchical top-down methodology

Chat with Paper

AI Agents for this Paper

Most frequently asked questions

1. What is the way to extract the relevant nodes?

2. What is the DTW similarity measure used to compare two histograms?

3. What is the way to learn the user’s behaviour from a segmentation example?

4. How many operations and time to process a HSR image?

5. How many operations are required to compute the DTW similarity measure?

6. Why is the aim of this article not to get the elongation results?

7. What is the problem with the standard distance used to compare two histograms?

8. How can the authors detect complex patterns without the noise induced by the details?

9. What is the way to deal with the variability in the image?

10. What is the histogram of The authorassociated to a subset of X ?

11. What are the main reasons why the proposed method cannot be compared?

12. Why is it important to develop tools to extract complex patterns from VHSR images?

Figures

Citations

Unmanned Aerial Vehicle for Remote Sensing Applications—A Review

Carotenoid content estimation in a heterogeneous conifer forest using narrow-band indices and PROSPECT + DART simulations

Review on Techniques for Plant Leaf Classification and Recognition

Very high resolution remote sensing image classification with SEEDS-CNN and scale effect analysis for superpixel CNN classification

Marker-Controlled Watershed-Based Segmentation of Multiresolution Remote Sensing Images

References

A theory for multiresolution signal decomposition: the wavelet representation

Normalized cuts and image segmentation

Normalized cuts and image segmentation

A review of assessing the accuracy of classifications of remotely sensed data

Dynamic programming algorithm optimization for spoken word recognition

Related Papers (5)

Object based image analysis for remote sensing

Geographic Object-Based Image Analysis - Towards a new paradigm.

Binary partition tree as an efficient representation for image processing, segmentation, and information retrieval

Mean shift: a robust approach toward feature space analysis

Antiextensive connected operators for image and sequence processing