Low-level aspects of segmentation and recognition

1992 ◽  
Vol 337 (1281) ◽  
pp. 371-379 ◽  
Keyword(s):  
Three Dimensional ◽  
Contour Length ◽  
Two Dimensional ◽  
Linear Combinations ◽  
Low Level ◽  
Parallel Network ◽  
Three Dimensional Models ◽  
Dimensional Object ◽  
Candidate Object ◽  
Selection Of

This paper discusses two problems related to three-dimensional object recognition. The first is segmentation and the selection of a candidate object in the image, the second is the recognition of a three-dimensional object from different viewing positions. Regarding segmentation, it is shown how globally salient structures can be extracted from a contour image based on geometrical attributes, including smoothness and contour length. This computation is performed by a parallel network of locally connected neuron-like elements. With respect to the effect of viewing, it is shown how the problem can be overcome by using the linear combinations of a small number of two-dimensional object views. In both problems the emphasis is on methods that are relatively low level in nature. Segmentation is performed using a bottom -up process, driven by the geometry of image contours. Recognition is performed without using explicit three-dimensional models, but by the direct manipulation of two-dimensional images.

Urban network of China from the perspective of population mobility: Three-dimensional co-occurrence of nodes and links

2021 ◽  
pp. 0308518X2199781
Author(s):  
Xinyue Luo ◽  
Mingxing Chen
Keyword(s):  
Three Dimensional ◽  
Geographical Distance ◽  
Two Dimensional ◽  
Population Mobility ◽  
Urban Network ◽  
Urban Networks ◽  
Low Level ◽  
Medium Level ◽  
High Level ◽  
External Connections

The nodes and links in urban networks are usually presented in a two-dimensional(2D) view. The co-occurrence of nodes and links can also be realized from a three-dimensional(3D) perspective to make the characteristics of urban network more intuitively revealed. Our result shows that the external connections of high-level cities are mainly affected by the level of cities(nodes) and less affected by geographical distance, while medium-level cities are affected by the interaction of the level of cities(nodes) and geographical distance. The external connections of low-level cities are greatly restricted by geographical distance.

Sight distance red zones on combined horizontal and sag vertical curves

1998 ◽  
Vol 25 (4) ◽  
pp. 621-630 ◽  
Author(s):  
Yasser Hassan ◽  
Said M Easa
Keyword(s):  
Quantitative Analysis ◽  
Three Dimensional ◽  
The Other ◽  
Two Dimensional ◽  
Horizontal Curve ◽  
Sight Distance ◽  
Stopping Sight Distance ◽  
Three Dimensional Models ◽  
Highway Alignment ◽  
Current Standards

Coordination of highway horizontal and vertical alignments is based on subjective guidelines in current standards. This paper presents a quantitative analysis of coordinating horizontal and sag vertical curves that are designed using two-dimensional standards. The locations where a horizontal curve should not be positioned relative to a sag vertical curve (called red zones) are identified. In the red zone, the available sight distance (computed using three-dimensional models) is less than the required sight distance. Two types of red zones, based on stopping sight distance (SSD) and preview sight distance (PVSD), are examined. The SSD red zone corresponds to the locations where an overlap between a horizontal curve and a sag vertical curve should be avoided because the three-dimensional sight distance will be less than the required SSD. The PVSD red zone corresponds to the locations where a horizontal curve should not start because drivers will not be able to perceive it and safely react to it. The SSD red zones exist for practical highway alignment parameters, and therefore designers should check the alignments for potential SSD red zones. The range of SSD red zones was found to depend on the different alignment parameters, especially the superelevation rate. On the other hand, the results showed that the PVSD red zones exist only for large values of the required PVSD, and therefore this type of red zones is not critical. This paper should be of particular interest to the highway designers and professionals concerned with highway safety.Key words: sight distance, red zone, combined alignment.

scholarly journals TSUNAMI GENERATION AND PROPAGATION

1972 ◽  
Vol 1 (13) ◽  
pp. 146
Author(s):  
Joseph L. Hammack ◽  
Frederic Raichlen
Keyword(s):  
Linear Theory ◽  
Source Region ◽  
Three Dimensional ◽  
Frequency Dispersion ◽  
Nonlinear Effects ◽  
Its Region ◽  
Experimental Results ◽  
Specific Class ◽  
Two Dimensional ◽  
Three Dimensional Models

A linear theory is presented for waves generated by an arbitrary bed deformation {in space and time) for a two-dimensional and a three -dimensional fluid domain of uniform depth. The resulting wave profile near the source is computed for both the two and three-dimensional models for a specific class of bed deformations; experimental results are presented for the two-dimensional model. The growth of nonlinear effects during wave propagation in an ocean of uniform depth and the corresponding limitations of the linear theory are investigated. A strategy is presented for determining wave behavior at large distances from the source where linear and nonlinear effects are of equal magnitude. The strategy is based on a matching technique which employs the linear theory in its region of applicability and an equation similar to that of Korteweg and deVries (KdV) in the region where nonlinearities are equal in magnitude to frequency dispersion. Comparison of the theoretical computations with the experimental results indicates that an equation of the KdV type is the proper model of wave behavior at large distances from the source region.

scholarly journals Optimization of the multidimensional signal interpolator in a lower dimensional space

2019 ◽  
Vol 43 (4) ◽  
pp. 653-660 ◽  
Author(s):  
M.V. Gashnikov
Keyword(s):  
Experimental Study ◽  
Dimensional Space ◽  
Three Dimensional ◽  
Two Dimensional ◽  
Multidimensional Signals ◽  
Multidimensional Signal ◽  
Signal Sample ◽  
Lower Dimensional Space ◽  
Lower Dimensional ◽  
Selection Of

Adaptive multidimensional signal interpolators are developed. These interpolators take into account the presence and direction of boundaries of flat signal regions in each local neighborhood based on the automatic selection of the interpolating function for each signal sample. The selection of the interpolating function is performed by a parameterized rule, which is optimized in a parametric lower dimensional space. The dimension reduction is performed using rank filtering of local differences in the neighborhood of each signal sample. The interpolating functions of adaptive interpolators are written for the multidimensional, three-dimensional and two-dimensional cases. The use of adaptive interpolators in the problem of compression of multidimensional signals is also considered. Results of an experimental study of adaptive interpolators for real multidimensional signals of various types are presented.

Three-Dimensional Mesh Model Watermarking Algorithm

2012 ◽  
Vol 198-199 ◽  
pp. 1481-1486
Author(s):  
Xin Li ◽  
Yi Ping Tian
Keyword(s):  
Wavelet Transform ◽  
Copyright Protection ◽  
Control Point ◽  
Three Dimensional ◽  
Digital Watermark ◽  
Two Dimensional ◽  
Mesh Model ◽  
Watermark Embedding ◽  
Three Dimensional Models ◽  
Watermark Algorithm

Watermark information is embedded in three-dimensional mesh model through three-dimensional watermarking algorithm for effective copyright protection. The widely use of three-dimensional grid model attracts more attention on the copyright protection. The digital watermark algorithm with the NURBS model based on the wavelet transform aims to get the virtual grayscale images using the control point coordinate. Then we can embed the watermark into the virtual gray image watermark. It can change the three-dimensional models into two-dimensional images. And this algorithm can enhance the operability and simplicity of the watermark embedding. Experiments show that the proposed algorithm is easy to implement, simple in principle, and the extracted watermark is clearly visible, moreover, the model does not need to be directly modified, so it has good robustness. Watermarked model does not change in the visual, it has good invisibility.

Two-Dimensional R-Matrices — Descendants of Three-Dimensional R-Matrices

1997 ◽  
Vol 12 (19) ◽  
pp. 1393-1410 ◽  
Author(s):  
S. M. Sergeev
Keyword(s):  
Three Dimensional ◽  
Baxter Equation ◽  
Two Dimensional ◽  
Affine Algebra ◽  
Tetrahedron Equation ◽  
Drinfeld Double ◽  
R Matrix ◽  
Operator Solution ◽  
Dimensional Models ◽  
Three Dimensional Models

Finite layers of three-dimensional models can be regarded as two-dimensional with complicated multi-stated weights. The tetrahedron equation in 3D provides the Yang–Baxter equation for this composite weights in 2D. Such solutions of the Yang–Baxter equation are constructed for the simplest operator solution of the tetrahedron equation. These R-matrices can be regarded as a special projection of universal R-matrix for some Drinfeld double [Formula: see text], associated with the affine algebra [Formula: see text]. Usual R-matrix for [Formula: see text] is another projection of [Formula: see text].

Statistical Approach to Shape from Shading: Reconstruction of Three-Dimensional Face Surfaces from Single Two-Dimensional Images

1996 ◽  
Vol 8 (6) ◽  
pp. 1321-1340 ◽  
Author(s):  
Joseph J. Atick ◽  
Paul A. Griffin ◽  
A. Norman Redlich
Keyword(s):  
Dimensional Space ◽  
Three Dimensional ◽  
Practical Interest ◽  
Principal Component ◽  
Shape From Shading ◽  
Shape Space ◽  
Two Dimensional ◽  
Two Dimensional Image ◽  
Low Dimensional ◽  
Dimensional Object

The human visual system is proficient in perceiving three-dimensional shape from the shading patterns in a two-dimensional image. How it does this is not well understood and continues to be a question of fundamental and practical interest. In this paper we present a new quantitative approach to shape-from-shading that may provide some answers. We suggest that the brain, through evolution or prior experience, has discovered that objects can be classified into lower-dimensional object-classes as to their shape. Extraction of shape from shading is then equivalent to the much simpler problem of parameter estimation in a low-dimensional space. We carry out this proposal for an important class of three-dimensional (3D) objects: human heads. From an ensemble of several hundred laser-scanned 3D heads, we use principal component analysis to derive a low-dimensional parameterization of head shape space. An algorithm for solving shape-from-shading using this representation is presented. It works well even on real images where it is able to recover the 3D surface for a given person, maintaining facial detail and identity, from a single 2D image of his face. This algorithm has applications in face recognition and animation.

Sign in / Sign up

Export Citation Format

Share Document