2020
DOI: 10.1007/978-3-030-40605-9_24
|View full text |Cite
|
Sign up to set email alerts
|

Dynamic Texture Representation Based on Hierarchical Local Patterns

Abstract: A novel effective operator, named HIerarchical LOcal Pattern (HILOP), is proposed to efficiently exploit relationships of local neighbors at each adjacent pairwise of different regional hierarchies located surrounding a center pixel of a texture image. Instead of thresholding by the value of central pixel, the gray-scale of each local neighbor in a hierarchical area is compared to that of all of neighbors in the remain region. In order to capture shape and motion cues for dynamic texture (DT) representation, H… Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
2
1
1
1

Citation Types

0
2
0

Year Published

2020
2020
2024
2024

Publication Types

Select...
8

Relationship

1
7

Authors

Journals

citations
Cited by 12 publications
(5 citation statements)
references
References 42 publications
0
2
0
Order By: Relevance
“…The proposed method achieves a rate of 99.50% on UCLA 50-class breakdown with LOO scheme. It is on par with MBSIF-TOP [16], MPCAF-TOP [20], FoSIG [63], V-BIG [64], CSAP-TOP [72], HILOP [73], B3DF_SMC [22], and ICFV [24]. Among these methods, MPCAF-TOP is a filter-learning-based multi-scale descriptor and ICFV involves two learning processes of filter learning and codebook construction, while our method is learning-free and thus has no dependence on data.…”
Section: ) Results On the Ucla Datasetmentioning
confidence: 96%
See 2 more Smart Citations
“…The proposed method achieves a rate of 99.50% on UCLA 50-class breakdown with LOO scheme. It is on par with MBSIF-TOP [16], MPCAF-TOP [20], FoSIG [63], V-BIG [64], CSAP-TOP [72], HILOP [73], B3DF_SMC [22], and ICFV [24]. Among these methods, MPCAF-TOP is a filter-learning-based multi-scale descriptor and ICFV involves two learning processes of filter learning and codebook construction, while our method is learning-free and thus has no dependence on data.…”
Section: ) Results On the Ucla Datasetmentioning
confidence: 96%
“…A relatively good result of 95.18% is provided by our DBRF on this dataset. It is slightly outperformed by ASF-TOP [17], MBSIF-TOP [16], MPCAF-TOP [20], FoSIG [63], V-BIG [64], HoGF 2D [25], HoGF 3D [25], DoDGF 2D [65], DoDGF 3D [65], DDLBP [66], novel LBP [18], MEWLSP [78], HILOP [73], MEMDP [76], RUBIG [74], B3DF_SMC [22], ICFV [24], and DT-RNNs [91] by 0.22%, 1.99%, 1.34%, 0.81%, 1.47%, 2.01%, 2.45%, 1.96%, 2.34%, 0.62%, 1.1%, 3.3%, 1.03%, 0.85%, 1.9%, 0.4%, and 1.33%, respectively. Despite the fact that many of these methods use SVM classifier, we think it is unworthy to significantly increase complexity and feature dimensionality for marginal performance improvement.…”
Section: ) Results On the Dyntex++ Datasetmentioning
confidence: 99%
See 1 more Smart Citation
“…The dynamic texture segmentation system was proposed by Paygude and Vyas [14] based on local and global spatiotemporal techniques. Dynamic texture representation was proposed by Nguyen et al [15] based on hierarchical local patterns.…”
Section: System Overviewmentioning
confidence: 99%
“…[20] nominated Local Binary Pattern (LBP) [21] for analyzing videos in two ways: VLBP patterns, which are structured from three consecutive frames of a sequence, and LBP-TOP patterns from three orthogonal planes. After that, some works proposed various schemes to enhance the distinguishing power by dealing with limitations of the typical LBPs in DT encoding such as problems of rotation-invariant [22], near-uniform regions, and sensitivity to noise [23,24,25,26].…”
Section: Introductionmentioning
confidence: 99%