Multiscale Fourier descriptor based on triangular features for shape retrieval

Yang, Chengzhuan; Yu, Qian

doi:10.1016/j.image.2018.11.004

Cited by 40 publications

(29 citation statements)

References 30 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The classification accuracy of our approach is 5% higher than the advanced exemplarbased method of Bioinformatics [33]. We also see that the recognition accuracy of our method is 21% higher than that of the multiscale Fourier descriptor [45]. Although our midlevel shape feature is built based on the multiscale Fourier descriptor, the high-level semantic information of an object's shape can be obtained using the FV encoding method.…”

Section: Methodsmentioning

confidence: 83%

“…Classification accuracy Class segment set [27] 69.70% IDSC [2] 73.60% Bag of SIFT [52] 74.90% Contour segments [29] 71.70% Skeleton paths [29] 67.90% ICS [29] 78.40% Multiscale Fourier descriptor [45] 67.35% Distinctive Parts [53] 80.70% BCF [22] 83.40 ± 1.30% Bioinformatics [33] 83.70% pSGLD [34] 84.80 ± 0.35% Contextual BOW [23] 86.00% BoSCP-LP [37] 89.77 ± 0.65% Our Method…”

Section: Methodsmentioning

confidence: 99%

“…Half training Leave-one-out Class segment set [27] 90.9% 97.93% IDSC+DP [27] 88.43% 90.71% Contour segments [29] 91.10% -Skeleton paths [29] 86.70% -ICS [29] 96.60% -HGTT [35] -95.09% Manifold kernel SVM [54] -96.57% Polygonal multi-resolution [55] -97.57% String of symbols [56] -97.36% Robust symbolic [28] -98.57% Kernel-edit distance [36] -98.93% BCF [22] 97.16% 98.93% Bioinformatics [33] 96.10% 98.10% BoSCP-LP [37] 98.72 ± 0.42% -Multiscale Fourier descriptor [45] 96.27% 97.93% Our Method 98.77 ± 0.65%…”

Section: Methodsmentioning

confidence: 99%

“…The triangular feature description of the contour sampling points introduced in this subsection is based on our previous work [45]. But unlike the method in [45], we do not need to use the final multiscale Fourier descriptor which is obtained using the Fourier transform. We only use the triangular features of the shape contour sampling points.…”

Section: A Definition Of Triangular Feature Representationmentioning

confidence: 99%

See 3 more Smart Citations

Learning Contour-Based Mid-Level Representation for Shape Classification

2020

Self Cite

View full text Add to dashboard Cite

This article proposes a novel contour-based mid-level shape description method for shape classification. This method resolves the shortcomings of low-level shape descriptors in dealing with the shapes of objects with large intra-class changes and non-linear deformation (articulation, occlusion and noise), thus improving the accuracy of shape classification. First, we extract the outer contour of an object and sample it. We next describe each sampling point on the shape contour with a triangular feature and regard it as a local feature. Then, a shape codebook is learned, and the Fisher vector encoding method is used to produce a compact mid-level shape feature. Finally, the learned mid-level shape features are sent to the linear support vector machine (SVM) classifier for shape recognition. The proposed method has been extensively tested on several standard shape datasets, and the experimental results show that our approach attains high accuracy of shape classification. Comparisons to other state-of-the-art shape classification approaches further prove the superiority and effectiveness of our method. INDEX TERMS Shape classification, triangular feature, fisher vector, mid-level shape representation.

show abstract

Section: Methodsmentioning

confidence: 83%

Section: Methodsmentioning

confidence: 99%

Section: Methodsmentioning

confidence: 99%

Section: A Definition Of Triangular Feature Representationmentioning

confidence: 99%

See 2 more Smart Citations

Learning Contour-Based Mid-Level Representation for Shape Classification

2020

Self Cite

View full text Add to dashboard Cite

show abstract

“…Although general purpose invariant descriptors are difficult to find, they can be effectively combined with other methods for shape-based object recognition, classification and indexing (Ullman, 1995). Shape features are stable image characteristics compared with colour and texture features and are relevant in the field of object recognition since they do not significantly change when the contrast and brightness are modified (Yang and Yu, 2019).…”

Section: Introductionmentioning

confidence: 99%

A multiscale shape descriptor based on diferential entropy

Pinheiro

Lopes

Carneiro

et al. 2019

Anais Do 14º Simpósio Brasileiro De Automação Inteligente

View full text Add to dashboard Cite

We propose a shape descriptor based on the differential entropy of multiscale curvature (MEC), which has a better shape discrimination ability than the normalised multiscale bending energy (NMBE) descriptor. We compare both NMBE and MEC with a monoscale descriptor by conducting multiple-class classification experiments on images from three public datasets: Kimia99 (99 shapes/9 classes), MPEG7-CE (1400 shapes/70 classes) and Flavia leaves (1907 shapes/32 classes). The classification results for precision, recall and F1-score measures, show that the use of MEC improved shape description for the MPEG7-CE and Flavia datasets. A qualitative analysis based on visualisation of pairwise Euclidean distance matrices also confirmed that MEC is a competitive method of shape description compared to NMBE for these datasets.

show abstract