Filters








57,453 Hits in 5.7 sec

Shape Matching and Object Recognition Using Low Distortion Correspondences

A.C. Berg, T.L. Berg, J. Malik
2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR'05)  
Face Recognition algorithm)Conclusion• Deformable shape matching.• Solving Correspondence problem • Geometric Blur concept • IQP and fast reduction usage • 300% improvement in Recognition rate(16% to  ...  Introduction • Recognition is a most important and interesting problem of Computer Vision. • Any recognition problem is fundamentally a problem of deformable shape matching. • According to the literature  ...  Assignment (Using IQP) Thank you for the patience.  ... 
doi:10.1109/cvpr.2005.320 dblp:conf/cvpr/BergBM05 fatcat:sd2fyhxfdbh63k63foizgnsxfm

Efficient shape matching using shape contexts

G. Mori, S. Belongie, J. Malik
2005 IEEE Transactions on Pattern Analysis and Machine Intelligence  
We present two algorithms for rapid shape retrieval: representative shape contexts, performing comparisons based on a small number of shape contexts, and shapemes, using vector quantization in the space  ...  We demonstrate that shape contexts can be used to quickly prune a search for similar shapes.  ...  We turn now to the use of shape contexts as part of a theory of object recognition based on shape matching.  ... 
doi:10.1109/tpami.2005.220 pmid:16285381 fatcat:sqg3huhqm5dv7hsx44lbjyxs6y

Computationally Efficient Graph Matching via Energy Vector Extraction [chapter]

Ariel Amato, Murad Al Haj, Josep Lládos, Jordi Gonzàlez
2007 Advances in Pattern Recognition  
Once these cues are extracted and quantified, a new energy function is used to match the different graphs based on the obtained features values.  ...  As a result the error of matching is reduced, as well as the computational cost by efficiently selecting and grouping representative features.  ...  Gold and Rangarajan [1] propose a one-to-one correspondence is established through a graduated assignment Kim and Kak [5] propose a bipartite graph matching using discrete relaxation for 3-D object  ... 
doi:10.1007/978-1-84628-945-3_5 fatcat:7zuu6jy32rhuribn3xljrabjry

Deformation-Driven Shape Correspondence

H. Zhang, A. Sheffer, D. Cohen-Or, Q. Zhou, O. van Kaick, A. Tagliasacchi
2008 Computer graphics forum (Print)  
Non-rigid 3D shape correspondence is a fundamental and difficult problem. Most applications which require a correspondence rely on manually selected markers.  ...  We present an automatic feature correspondence algorithm capable of handling large, non-rigid shape variations, as well as partial matching.  ...  Using shape extremities as features also reinforces the correlation between correspondence analysis and object recognition by parts.  ... 
doi:10.1111/j.1467-8659.2008.01283.x fatcat:ayojawdzxjgxtfc36lgkadbyy4

Beyond one-to-one feature correspondence: The need for many-to-many matching and image abstraction

S. Dickinson
2009 2009 IEEE Computer Society Conference on Computer Vision and Pattern Recognition Workshops  
Over the last four decades, object recognition systems have typically assumed a one-to-one correspondence between image and model features.  ...  In this talk, I will briefly review three formulations of the many-to-many matching problem as applied to model acquisition, model indexing, and object recognition.  ...  Drawing on low-distortion graph embedding techniques, we first embed the nodes of two graphs to be matched into a lowdimensional vector space, such that attributed nodes map to weighted points and shortest-path  ... 
doi:10.1109/cvpr.2009.5204333 fatcat:rsdkltqosfcbtltsxsube5e2jq

Beyond one-to-one feature correspondence: The need for many-to-many matching and image abstraction

Sven Dickinson
2009 2009 IEEE Computer Society Conference on Computer Vision and Pattern Recognition Workshops  
Over the last four decades, object recognition systems have typically assumed a one-to-one correspondence between image and model features.  ...  In this talk, I will briefly review three formulations of the many-to-many matching problem as applied to model acquisition, model indexing, and object recognition.  ...  Drawing on low-distortion graph embedding techniques, we first embed the nodes of two graphs to be matched into a lowdimensional vector space, such that attributed nodes map to weighted points and shortest-path  ... 
doi:10.1109/cvprw.2009.5204333 dblp:conf/cvpr/Dickinson09 fatcat:mwqfig2psfeyznxav5mmh5hgju

Object Categorization and the Need for Many-to-Many Matching [chapter]

Sven Dickinson, Ali Shokoufandeh, Yakov Keselman, Fatih Demirci, Diego Macrini
2005 Lecture Notes in Computer Science  
We review three formulations of the many-to-many matching problem as applied to model acquisition and object recognition.  ...  We review the evolution of object recognition systems and argue that bridging this representational gap requires an ability to match image and model features manyto-many.  ...  Two graphs to be matched are embedded with low distortion into a low-dimensional Euclidean space, the two weighted sets of points are matched many-to-many using the Earth Mover's Distance (EMD) algorithm  ... 
doi:10.1007/11550518_62 fatcat:b3ybc4jqdjbjnnocdmzn6f6erm

Object Recognition as Many-to-Many Feature Matching

M. Fatih Demirci, Ali Shokoufandeh, Yakov Keselman, Lars Bretzner, Sven Dickinson
2006 International Journal of Computer Vision  
Object recognition can be formulated as matching image features to model features. When recognition is exemplar-based, feature correspondence is one-to-one.  ...  This is accomplished by embedding the initial weighted graphs into a normed vector space with low distortion using a novel embedding technique based on a spherical encoding of graph structure.  ...  deformation and articulation, and minor within-class shape deformation.  ... 
doi:10.1007/s11263-006-6993-y fatcat:35o2peidqbcoxo7tfnciluu7lm

Efficient many-to-many feature matching under the l1 norm

M. Fatih Demirci, Yusuf Osmanlioglu, Ali Shokoufandeh, Sven Dickinson
2011 Computer Vision and Image Understanding  
Matching configurations of image features, represented as attributed graphs, to configurations of model features is an important component in many object recognition algorithms.  ...  The Earth Movers Distance (EMD) algorithm was then used to match the resulting points, with the computed flows specifying the many-to-many vertex correspondences between the input graphs.  ...  Acknowledgments The work of Fatih Demirci is supported in part by The Scientific and Technological Research Council of Turkey (TÜB _ ITAK) Grant 109E183.  ... 
doi:10.1016/j.cviu.2010.12.012 fatcat:mkpef2uq2vcyzffuglfiqs7tey

Pictorial recognition of objects employing affine invariance in the frequency domain

J. Ben-Arie, Zhiqian Wang
1998 IEEE Transactions on Pattern Analysis and Machine Intelligence  
This paper describes an e cient approach to pose invariant pictorial object recognition employing spectral signatures of image patches that correspond to object surfaces which are roughly planar.  ...  The images are preprocessed by a novel model based segmentation scheme that detects and segments objects that are a ne-similar to members of a model set of basic geometric shapes.  ...  Recognition of Multiple Patches and Non-Isolated Objects Using Model Based Segmentation In eigenspace matching).  ... 
doi:10.1109/34.683774 fatcat:pixmvtdo4zgahdvqorjbkffi34

Recognizing Partially Occluded Object from a Line Drawing

Guimei Zhang, Jun Chu
2011 Journal of Computers  
Moreover, they establish one-to-one correspondence between model features and object features in a scene, and can recognize multiple objects.  ...  Most existing methods for partially occluded object recognition are only suit for Euclid and similarity transformation.  ...  When one object occlude the other, the distortion of an isolated region of the shape will result in changes to every feature.  ... 
doi:10.4304/jcp.6.8.1740-1747 fatcat:6t5zb2z7gzcvxpizlx2wj6ai7a

Shape Matching and Object Recognition [chapter]

Alexander C. Berg, Jitendra Malik
2006 Lecture Notes in Computer Science  
Shape Matching and Object Recognition We address comparing related, but not identical shapes in images following a deformable template strategy.  ...  The transformation necessary for alignment and the remaining differences after alignment are then used to make a comparison.  ...  We break the problem of matching shapes into computing measures of local shape similarity and finding a low distortion correspondence between keypoints that have similar local shapes.  ... 
doi:10.1007/11957959_25 fatcat:4vymq73bobcmpesw7ipaamj2ge

Automatic Target Recognition Based on Cross-Plot

Kelvin Kian Loong Wong, Derek Abbott, Teresa Serrano-Gotarredona
2011 PLoS ONE  
Target recognition was implemented based on the proposed pattern recognition concept and tested rigorously for its precision and recall performance.  ...  To achieve this objective, Cross-plots of binary patterns are explored as potential signatures for the observed target by high-speed capture of the crucial spatial features using minimal computational  ...  Successful retrieval systems have good matching abilities for shape objects that are subjected to distortion, scaling, translation, noise, and region loss when using the same feature set.  ... 
doi:10.1371/journal.pone.0025621 pmid:21980508 pmcid:PMC3183066 fatcat:4ln5wqv3tzdgrimiskdqmd4i2q

Gabor wavelet representation for 3-D object recognition

Xing Wu, Bir Bhanu
1997 IEEE Transactions on Image Processing  
This paper presents a model-based object recognition approach that uses a Gabor wavelet representation.  ...  The key idea is to use magnitude, phase, and frequency measures of the Gabor wavelet representation in an innovative flexible matching approach that can provide robust recognition.  ...  The confusion matrix is given in Table V and the recognition performance is 76%. The reason for the errors is significant distortion in shape and signature variations.  ... 
doi:10.1109/83.552096 pmid:18282878 fatcat:k42bfpicujevzgkzi4zwu4ksgq

A Biological Inspired Visual Landmark Recognition Architecture

Quoc Do, Lakhmi Jain
2009 2009 Digital Image Computing: Techniques and Applications  
and landmark recognition.  ...  An architecture that is inspired by a human's capability to autonomously navigate an environment based on visual landmark recognition is presented.  ...  The concept of image diffusion and shape attraction processes to achieve distortion and small size and shape invariant recognition is further illustrated in Figure 5 .  ... 
doi:10.1109/dicta.2009.61 dblp:conf/dicta/DoJ09 fatcat:cfpi4cwdgzfbfc4rncwa6spq5m
« Previous Showing results 1 — 15 out of 57,453 results