Encoding Spatial Arrangements of Visual Words for Rotation-Invariant Image Classification

Anwar, Hafeez; Zambanini, Sebastian; Kampel, Martin

doi:10.1007/978-3-319-11752-2_36

Cited by 13 publications

(17 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We previously proposed [23] to use all the distinct pairs of three descriptors from set Di to calculate angles between the spatial positions of the descriptors as shown in Figure 6. We call that method combinatorial triangulation, as the triangulation is done for all the three distinct pairs of descriptors belonging to a given visual word.…”

Section: [Fig6] Triangulation Methods (A) Delaunay Triangulation (Bmentioning

confidence: 99%

“…Second, we apply the circular tiling scheme [20] over the segmented coin image to increase the discriminative power of the model while maintaining rotation invariance. Finally, the rotation-invariant geometric relationships of visual words [23], [24] in each subregion of the circular tiling scheme are modeled to obtain the final image representation. In the following, we will give a brief description of each step of the proposed strategy.…”

Section: Rotation-invariant Spatial Extensions To the Bow Image Reprementioning

confidence: 99%

See 1 more Smart Citation

Ancient Coin Classification Using Reverse Motif Recognition: Image-based classification of Roman Republican coins

Anwar

Zambanini

Kampel

et al. 2015

IEEE Signal Process. Mag.

Self Cite

View full text Add to dashboard Cite

e propose a holistic system to classify ancient Roman Republican coins based on their reverse-side motifs. The bag-ofvisual-words (BoW) model is enriched with spatial information to increase the discriminative power of the coin image representation. This is achieved by combining a spatial pooling scheme with co-occurrence encoding of visual words. We specifically address the required geometric invariance properties of image-based ancient coin classification, as coins from different collections can be located at differing image locations, have various scales in the images computer engineering from Myongji University, South Korea, where he also acted as a research assistant on a funded project related to the design and development of an intra-oral threedimensional scanner. He is currently pursuing his Ph.D. degree at the Computer Vision

show abstract

Section: [Fig6] Triangulation Methods (A) Delaunay Triangulation (Bmentioning

confidence: 99%

Section: Rotation-invariant Spatial Extensions To the Bow Image Reprementioning

confidence: 99%

Ancient Coin Classification Using Reverse Motif Recognition: Image-based classification of Roman Republican coins

Anwar

Zambanini

Kampel

et al. 2015

IEEE Signal Process. Mag.

Self Cite

View full text Add to dashboard Cite

show abstract

“…This allows to use the local features from the foreground to construct the visual vocabulary, thus making it more discriminating and accurate as shown in [1].…”

Section: Methodsmentioning

confidence: 99%

Invariant Image-Based Species Classification of Butterflies and Reef Fish

Anwar¹,

Zambanini²,

Kampel³

2015

Procedings of the Machine Vision of Animals and Their Behaviour Workshop 2015

Self Cite

View full text Add to dashboard Cite

We propose a framework for species-based image classification of butterflies and reef fish. To support such image-based classification, we use an image representation which enriches the famous bag-of-visual words (BoVWs) model with spatial information. This image representation is developed by encoding the global geometric relationships of visual words in the 2D image plane in a scale-and rotation-invariant manner. In this way, invariance is achieved to the most common variations found in the images of these animals as they can be imaged at different image locations, exhibit various in-plane orientations and have various scales in the images. The images in our butterfly and reef fish datasets belong to 30 species of each animal. We achieve better classification rates on both the datasets than the ordinary BoVWs model while still being invariant to the mentioned image variations. Our proposed image-based classification framework for butterfly and reef fish species can be considered as a helpful tool for scientific research, conversation and education.

show abstract

“…(2) We previously proposed [17] to use all the distinct pairs of three descriptors from set to calculate angles between the spatial positions of the descriptors as shown in Fig. 2.…”

Section: Scale-and Rotation-invariant Histogram Of Identical Visumentioning

confidence: 99%

“…Rotation-invariant SIFT features are extracted and concatenated at predefined scales of . We previously showed that the use of segmentation masks at the stage of vocabulary construction enhances the discriminating nature of the vocabulary, thus resulting in a higher classification rate [17]. Here we also use segmentation masks to extract the foreground features for vocabulary construction.…”

Section: A Number Of Scales For Local Features Extractionmentioning

confidence: 99%

Efficient Scale- and Rotation-Invariant Encoding of Visual Words for Image Classification

Anwar

Zambanini

Kampel

2015

IEEE Signal Process. Lett.

Self Cite

View full text Add to dashboard Cite

The problem of incorporating spatial information to the bag-of-visual-words model for image classification is addressed in this letter. To incorporate such information, we propose to encode the global geometric relationships of the visual words in the image plane in a scale-and rotation-invariant manner. This is established by measuring scale-and rotation-invariant geometrical properties given by triangles of identical visual words. Experimental results demonstrate that our proposed method is more robust to changes in scale and image rotations than the bag-of-visual words model on a butterfly and fish dataset.

show abstract

Encoding Spatial Arrangements of Visual Words for Rotation-Invariant Image Classification

Cited by 13 publications

References 12 publications

Ancient Coin Classification Using Reverse Motif Recognition: Image-based classification of Roman Republican coins

Ancient Coin Classification Using Reverse Motif Recognition: Image-based classification of Roman Republican coins

Invariant Image-Based Species Classification of Butterflies and Reef Fish

Efficient Scale- and Rotation-Invariant Encoding of Visual Words for Image Classification

Contact Info

Product

Resources

About