2017年CIARP大会论文集：模式识别、图像分析与计算机视觉的最新进展

5星 · 超过95%的资源需积分: 10 163 浏览量更新于2024-07-18 2 收藏 83.3MB PDF 举报

本论文集《Pattern Recognition, Image Analysis, Computer Vision》由Marcelo Mendoza和Sergio Velastín共同编辑，收录于第123卷的 LNCS（ Lecture Notes in Computer Science）系列，编号为10657。它是2017年Iberoamerican Congress, CIARP在智利瓦尔帕莱索举办的第22届会议上发表的成果汇编。会议时间定于11月7日至10日，专注于模式识别、图像分析以及计算机视觉领域的最新进展和应用。论文集强调了该领域自1973年开始的持续发展，其编委会成员包括多位国际知名学者，如David Hutchison（英国兰开斯特大学）、Takeo Kanade（美国卡内基梅隆大学）、Josef Kittler（英国萨里大学）、Jon Kleinberg（美国康奈尔大学）、Friedemann Mattern（瑞士苏黎世联邦理工学院）、John Mitchell（美国斯坦福大学）、Moni Naor（以色列魏茨曼科学研究所）、C. P. Pandu Rangan（印度马德拉斯印度理工学院）、Bernhard Steffen（德国多特蒙德大学）、Demetri Terzopoulos（美国加州大学洛杉矶分校）和Doug Tygar（美国加州大学伯克利分校），以及Gerhard Weikum（德国萨尔布吕肯马克斯·普朗克信息学研究所）等，他们的专业知识为本领域的研究提供了深厚背景和指导。该论文集的内容可能涵盖了多种技术主题，如特征提取与匹配、机器学习在图像分类中的应用、深度学习在计算机视觉中的突破、目标检测和跟踪、图像分割、图像增强与重建、视觉传感器融合、生物特征识别、自动驾驶中的视觉导航以及人工智能在艺术与文化分析中的角色。此外，书中还可能探讨了计算机视觉在医学成像、工业自动化、安防监控和虚拟现实等实际场景中的实践和挑战。整体来说，这是一本深入探讨当今计算机视觉核心技术与应用的重要文献，对于研究人员、工程师以及对这一领域感兴趣的学生来说，是理解和掌握最新研究成果及发展趋势的宝贵资源。通过阅读此论文集，读者可以了解到当前国际上在这个复杂而快速发展的领域中的前沿工作，并为自己的研究或实践提供有价值的参考。

On Semantic Solutions for Efficient Approximate Similarity Search

on Large-Scale Datasets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 450

Alexander Ocsa, Jose Luis Huillca, and Cristian Lopez del Alamo

Self-organizing Maps for Motor Tasks Recognition from Electrical

Brain Signals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 458

Alvaro D. Orjuela-Cañón, Osvaldo Renteria-Meza, Luis G. Hernández,

AndrésF.Ruí z-Olaya, Al exander Cerquera, and Javier M. Antelis

Semantic Segmentation of Color Eye Images for Improving

Iris Segmentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 466

Dailé Osorio-Roig, Annette Morales-González,

and Eduardo Garea-Llano

A Single-Step 2D Thinning Scheme with Deletion of P-Simple Points. . . . . . 475

Kálmán Palágyi and Péter Kardos

Non Local Means Image Filtering Using Clustering. . . . . . . . . . . . . . . . . . . 483

Alvaro Pardo

Neural Networks for the Reconstruction and Separation of High Energy

Particles in a Preshower Calorimeter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 491

Juan Pavez, Hayk Hakobyan, Carlos Valle, William Brooks,

Sergey Kuleshov, and Héctor Allende

Noisy Character Recognition Using Deep Convolutional

Neural Networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 499

Sirlene Peixoto, Gabriel Gonçalves, Andrea Bianchi,

Alceu De S. Brito, William Robson Schwartz, and David Menotti

A Flexible Framework for the Evaluation of Unsupervised

Image Annotation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 508

Luis Pellegrin, Hugo Jair Escalante, Manuel Montes-y-Gómez,

Mauricio Villegas, and Fabio A. González

Unsupervised Local Regressive Attributes for Pedestrian Re-identification . . . 517

Billy Peralta, Luis Caro, and Alvaro Soto

A Deep Boltzmann Machine-Based Approach for Robust

Image Denoising. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 525

Rafael G. Pires, Daniel S. Santos, Gustavo B. Souza,

Aparecido N. Marana, Alexandre L. M. Levada,

and João Paulo Papa

Discovering Bitcoin Mixing Using Anomaly Detection . . . . . . . . . . . . . . . . 534

Mario Alfonso Prado-Romero, Christian Doerr,

and André s Gago-Alonso

XVI Contents

A Scale-Space Approach for Multiscale Shape Analysis. . . . . . . . . . . . . . . . 542

Lucas Alexandre Ramos, Aparecido Nilceu Marana,

and Luis Antônio de Souza Junior

An Unsupervised Approach for Eye Sclera Segmentation . . . . . . . . . . . . . . . 550

Daniel Riccio, Nadia Brancati, Maria Frucci,

and Diego Gragnaniello

An Approach to Clustering Using the Expectation-Maximization

and Selection of Attributes ReliefF Applied to Water Treatment

Plants process. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 558

Fábio Cosme Rodrigues dos Santos, André Felipe Henriques Librantz,

and Renato José Sassi

A Two-Step Neural Network Approach to Passage Retrieval for Open

Domain Question Answering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 566

Andrés Rosso-Mateus, Fabio A. González, and Manuel Montes-y-Gómez

Ensembles of Multiobjective-Based Classifiers for Detection

of Epi leptic Seizures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 575

Fernando S. Beserra , Marcos M. Raimundo,

and Fernando J. Von Zuben

Class Confusability Reduction in Audio-Visual Speech Recognition

Using Random Forests. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 584

Gonzalo D. Sad, Lucas D. Terissi, and Juan C. Gómez

Exploring Image Bit Planes for Video Shot Boundary Detection . . . . . . . . . . 593

Anderson Carlos Sousa e Santos and Hel io Pedrini

Convolutional Network for EEG-Based Biometric . . . . . . . . . . . . . . . . . . . . 601

Thiago Schons, Gladston J. P. Moreira, Ped ro H. L. Silva,

Vitor N. Coelho, and Eduardo J. S. Luz

Skeleton Pruning Based on Elongation and Size of Object’s Limbs

and Boundary’s Convexities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 609

Luca Serino and Gabriella Sanniti di Baja

Volume Rendering by Stochastic Neighbor Embedding-Based 2D Transfer

Function Building . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 618

Walter Serna-Serna, Andres M. Álvarez-Meza,

and Álvaro-Ángel Orozco-Gutierrez

Text Localization in Born-D igital Images of Advertisements. . . . . . . . . . . . . 627

Dirk Siegmund, Aidmar Wainakh, Tina Ebert, Andreas Braun,

and Arjan Kuijper

Contents XVII

Deep Learning Techniques Applied to the Cattle Brand Recognition . . . . . . . 635

Carlos Silva, Daniel Welfer, and Juliano Weber

Efficient Transfer Learning for Robust Face Spoofing Detection . . . . . . . . . . 643

Gustavo B. Souza, Daniel F. S. Santos, Rafael G. Pires,

Aparecido N. Marana, and João P. Papa

A Kernel-Based Optimum-Path Forest Classifier . . . . . . . . . . . . . . . . . . . . . 652

Luis C. S. Afonso, Danillo R. Pereira, and João P. Papa

A Possibilistic c-means Clustering Model with Cluster Size Estimation . . . . . 661

László Szilágyi and Sándor M. Szilágyi

Randomized Neural Network Based Signature for Classification

of Titanium Alloy Microstructures. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 669

Jarbas Joaci de Mesquita Sá Junior, André R. Backes,

and Odemir Martinez Bruno

Pap-smear Image Classification Using Randomized Neural Network

Based Signature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 677

Jarbas Joaci de Mesquita Sá Junior, André R. Backes,

and Odemir Martinez Bruno

Estimation of Pedestrian Height Using Uncalibrated Cameras . . . . . . . . . . . . 685

Alejandro Valdés-Camejo, Guillermo Aguirre-Carrazana,

Raúl Alonso-Baryolo, Annette Morales-González,

Francisco J. Silva-Mata, and Edel García-Reyes

Semi-supervised Online Kernel Semantic Embedding

for Multi-label Annotation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 693

Jorge A. Vanegas, Hugo Jair Escalante, and Fabio A. González

Ultrasonic Backscatter Signal Processing Technique

for the Characte rization of Animal Lymph Node . . . . . . . . . . . . . . . . . . . . . 702

Julian A. Villamarín, Daniela A. Montilla, Olga M. Potosi,

Luis F. Londoño, Fabian G. Muñoz, and Edgar W. Gutierrez

A Distributed Shared Nearest Neighbors Clustering Algori thm . . . . . . . . . . . 710

Juan Zamora, Héctor Allende-Cid, and Marcelo Mendoza

Automatic Classification of Optical Defects of Mirrors from Ronchigram

Images Using Bag of Visual Words and Support Vector Machines . . . . . . . . 719

Daniel Zapata, Ang el Cruz-Roa, and Andrés Jiménez

Author Index ............................................ 727

XVIII Contents

A Citation k-NN Approach for Facial

Expression Recognition

Daniel Acevedo

1,2(

)

, Pablo Negri

,Mar´ıa Elena Buemi

Francisco G´omez Fern´andez

1,2

,andMartaMejail

1,2

Facultad de Ciencias Exactas y Naturales, Departamento de Computaci´on,

Universidad de Buenos Aires, Buenos Aires, Argentina

dacevedo@dc.uba.ar

Instituto de Investigaci´on en Ciencias de la Computaci´on (ICC),

CONICET-Universidad de Buenos Aires, Buenos Aires, Argentina

CONICET-Universidad Argentina de la Empresa (UADE),

Buenos Aires, Argentina

Abstract. The identiﬁcation of facial expressions with human emotions

plays a key role in non-verbal human communication and has applica-

tions in several areas. In this work, we propose a descriptor based on

areas and angles of triangles formed by the landmarks from face images.

We test this descriptors for facial expression recognition by means of an

adaptation of the k-Nearest Neighbors classiﬁer called Citation-kNN in

which the training examples come in the form of sets of feature vectors.

Comparisons with other state-of-the-art techniques on the CK+ dataset

are shown. The descriptor remains robust and precise in the recognition

of expressions.

1 Introduction

Facial expressions have been studied due to the great deal of applications that are

required for interpreting human communication through facial gestures, such as

automatic behavior recognition, human-computer interaction, pain assessment,

health-care, surveillance, deceit detection and sign language recognition [1]. In

Human Computer Interaction (HCI), it can be used to supersede other forms of

non-verbal communication, since the expressiveness of faces are usually linked

to an emotional state [2].

We aim at recognizing human emotions from the complete set of the Extended

Cohn-Kanade AU-Coded Facial Expression Database (CK+) [3]. This database

captures several persons performing facial expressions, including seven basic

emotions: anger, contempt, disgust, fear, happy, sadness and surprise.

In this work, a geometric descriptor that captures facial expression evolution

within a video sequence is introduced. It is based on the angles and areas formed

by the landmarks placed on the images of faces that perform an expression.

A similar descriptor was introduced in [4] and its dynamic was modeled with

a Conditional Random Field (as opposed to our Multiple Instance Learning

 Springer International Publishing AG, part of Springer Nature 2018

M. Mendoza and S. Velast´ın (Eds.): CIARP 2017, LNCS 10657, pp. 1–9, 2018.

https://doi.org/10.1007/978-3-319-75193-1

2 D. Acevedo et al.

approach presented in this work). Geometric descriptors were also proposed for

other applications and diﬀerent contexts. In [5] geometric descriptors are used for

kinship veriﬁcation. Angles from triangles formed by 76 landmarks are extracted

from images as well as segment lengths and its ratios. In [6] the authors propose

for age invariant face recognition the use of a set of triangles proportion ratios to

estimate the degree of similarity among triangles. A thorough study of diﬀerent

methodologies can be found at [7].

We use the geometric descriptor (angles and areas) in a variant of a supervised

machine learning framework in which the descriptors are grouped in a form of

bags. By means of a distance between bags (Hausdorﬀ distance), an adaptation of

the k-nearest neighbor algorithm is used as a classiﬁer, called Citation k-NN [8].

The paper is organized as follows: Sect. 2 introduces the proposed descriptor.

Section 3 presents the adaptation of our geometric descriptor to be used by the

Citation k-NN classiﬁer. Section 4 presents the experimental results. Finally

Sect. 5 presents conclusions.

2 Geometrical Descriptor

Expression recognition using temporal approaches usually tracks the changes of

landmarks spatial positions within the images on a video sequence. This analysis,

however, is sensible to movement of the head while the expression is occurring.

Our descriptor, computes a measure independent of the spatial position of

the landmarks on the image. They are computed from the triangles obtained

by three landmarks, inspired on the polygon mesh representation. In this way,

the approach is independent of the face pose, and the features only measure

facial transformations by evaluating the changes on the angles and areas on

the respective triangles between consecutive frames. Three landmarks (

,

)

deﬁne a triangle Δ. The geometry of Δ can be described by its area and the

internal angles.Thus,letbe:θ

, the angular descriptor where 

is the landmark

at the central point, 

and 

are the extreme points of the angle, and α

is the

area of triangle Δ. Figure 1 shows a subset of landmarks and the evolution of an

Fig. 1. Images of a surprise expression with landmarks numbered and the triangle

Δ =(

,

) displayed in red lines, (a) t

− 5 frame, centered at angle



=45

◦

(b) t

frame, centered at angle



=96

◦

剩余747页未读，继续阅读

J-10

粉丝: 18
资源: 484

2017年CIARP大会论文集：模式识别、图像分析与计算机视觉的最新进展

Progress in Pattern Recognition Image Analysis Computer Vision and Applications

computer vision: models learning and inference（计算机视觉by Simon J. D. Prince）

Pattern Recognition and Image Analysis

Progress in Pattern Recognition, Image Analysis, Computer Vision

Progress in Pattern Recognition, Image Analysis, Computer Vision, and App

Bishop Pattern Recognition and Machine Learning

Progress in Artificial Intelligence and Pattern Recognition

Moments and Moment Invariants in Pattern Recognition

computer-vision-notebooks::eye:Computer Vision基础知识和有用的python笔记本的权威收藏

最新资源