3D Geometric Scale Variability in Range Images: Features and Descriptors

Bariya, Prabin; Novatnack, John; Schwartz, Gabriel; Nishino, Ko

doi:10.1007/s11263-012-0526-7

Cited by 65 publications

(69 citation statements)

References 26 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…In effect, the average surface normal captured by a given pixel is computed across a greater area as scale increases. This observation, also underlying the recent work on geometric scale space of range images [1,18], motivates our approach to modeling the scale variability of geometric texture. We describe the scale space of geometric texture by filtering the surface normal field with Gaussian kernels of increasing standard deviation.…”

Section: Scale-space Representationmentioning

confidence: 62%

See 1 more Smart Citation

The Scale of Geometric Texture

Oxholm

Bariya

Nishino

2012

Computer Vision – ECCV 2012

Self Cite

View full text Add to dashboard Cite

Abstract. The most defining characteristic of texture is its underlying geometry. Although the appearance of texture is as dynamic as its illumination and viewing conditions, its geometry remains constant. In this work, we study the fundamental characteristic properties of texture geometry-self similarity and scale variability-and exploit them to perform surface normal estimation, and geometric texture classification. Textures, whether they are regular or stochastic, exhibit some form of repetition in their underlying geometry. We use this property to derive a photometric stereo method uniquely tailored to utilize the redundancy in geometric texture. Using basic observations about the scale variability of texture geometry, we derive a compact, rotation invariant, scale-space representation of geometric texture. To evaluate this representation we introduce an extensive new texture database that contains multiple distances as well as in-plane and outof plane rotations. The high accuracy of the classification results indicate the descriptive yet compact nature of our texture representation, and demonstrates the importance of geometric texture analysis, pointing the way towards improvements in appearance modeling and synthesis.

show abstract

Section: Scale-space Representationmentioning

confidence: 62%

“…8, we introduce a new database that covers 20 textures at different distances, with different in-plane and out-of-plane rotations 1 . To our knowledge, this is the only public database that offers multiple distances for each texture in addition to multiple in-plane and out-of-plane rotations.…”

Section: Geometric Texture Databasementioning

confidence: 99%

The Scale of Geometric Texture

Oxholm

Bariya

Nishino

2012

Computer Vision – ECCV 2012

Self Cite

View full text Add to dashboard Cite

show abstract

“…10 However, most of the existing feature descriptors still suffer from either low descriptiveness or weak robustness. 18 Wang et al 19 presented a sphere-spin-image (SSI) descriptor by mapping 3-D coordinates of points within a sphere centered at a key point into 2-D space, which is more descriptive than traditional SI. Inspired by the idea of USC 16 feature descriptor extending from 3-D shape context 20 (3-DSC), Guo et al 21 proposed the TriSI descriptor which was concatenated by three spin images generated using the x − y − z-axes as the spin axis.…”

Section: Introductionmentioning

confidence: 99%

Simple and efficient improvement of spin image for three-dimensional object recognition

Hao

2016

Opt. Eng

View full text Add to dashboard Cite

Abstract. This paper presents a highly distinctive and robust local three-dimensional (3-D) feature descriptor named longitude and latitude spin image (LLSI). The whole procedure has two modules: local reference frame (LRF) definition and LLSI feature description. We employ the same technique as Tombari to define the LRF. The LLSI feature descriptor is obtained by stitching the longitude and latitude (LL) image to the original spin image vertically, where the LL image was generated similarly with the spin image by mapping a two-tuple ðθ; ϕÞ into a discrete two-dimensional histogram. The performance of the proposed LLSI descriptor was rigorously tested on a number of popular and publicly available datasets. The results showed that our method is more robust with respect to noise and varying mesh resolution than existing techniques. Finally, we tested our LLSI-based algorithm for 3-D object recognition on two popular datasets. Our LLSI-based algorithm achieved recognition rates of 100%, 98.2%, and 96.2%, respectively, when tested on the Bologna, University of Western Australia (UWA) (up to 84% occlusion), UWA datasets (all). Moreover, our LLSI-based algorithm achieved 100% recognition rate on the whole UWA dataset when generating the LLSI descriptor with the LRF proposed by Guo.

show abstract

“…There are two basic types of features that are used to represent a range image, namely global features and local features. Global feature representations are widely used in shape retrieval and object detection techniques, but these features are sensitive to clutter and occlusion [1,12]. On the other hand, a local surface feature is robust to these conditions as it is more distinctive and descriptive.…”

Section: Introductionmentioning

confidence: 99%