Slezak / Pal / Kang Signal Processing, Image Processing and Pattern Recognition,

International Conference, SIP 2009, Held as Part of the Future Generation Information Technology Conference, FGIT 2009, Jeju Island, Korea, December 10-12, 2009. Proceedings
1. Auflage 2010
ISBN: 978-3-642-10546-3
Verlag: Springer
Format: PDF
Kopierschutz: 1 - PDF Watermark

E-Book, Englisch, Band 61, 341 Seiten

Reihe: Communications in Computer and Information Science

This book constitutes the proceedings of the International Conference on Signal Processing, Image Processing and Pattern Recognition, SIP 2009, held as part of the Future Generation Information Technology Conference, FGIT 2009, held on Jeju Island, Korea, December 10-12, 2009. The 38 papers presented in this volume were carefully reviewed and selected from numerous submissions. The topics covered are from multifaceted aspects of signal processing, image processing and pattern recognition.

Autoren/Hrsg.

Slezak, Dominik

Pal, Sankar

Kang, Byeong-Ho

Gu, Junzhong

Kuroda, Hideo

Kim, Tai-Hoon

Inhaltsverzeichnis

1;Title Page;2
2;Foreword;5
3;Preface;7
4;Organization;8
5;Table of Contents;9
6;A Blind Image Wavelet-Based Watermarking Using Interval Arithmetic;13
6.1;Introduction;13
6.2;Interval Arithmetic;14
6.3;Interval Wavelet Decomposition;14
6.4;Watermarking Algorithm;15
6.5;Considerations on the Proposed Algorithms;16
6.5.1;The Choice of Parameters;16
6.5.2;Relationship between Our Method and the HVS;17
6.6;Experimental Results;18
6.7;Conclusion;20
6.8;References;20
7;Hand Gesture Spotting Based on 3D Dynamic Features Using Hidden Markov Models;21
7.1;Introduction;21
7.2;Hidden Markov Models;22
7.3;3D Dynamic Feature Extraction;23
7.4;Gesture Spotting and Recognition System;24
7.5;Experimental Results;25
7.5.1;Isolated Gesture Recognition;25
7.5.2;Meaningful Gesture Spotting Test;26
7.6;Conclusion and Future Work;27
7.7;References;28
8;Objective Quality Evaluation of Laser Markings for Assembly Control;29
8.1;Introduction;29
8.2;Marking Evaluation;30
8.2.1;Marking Segmentation and Template Matching;30
8.2.2;Distortion Correction by Raster Alignment;31
8.2.3;Poll about Human Perception;32
8.2.4;Evaluation Methods;33
8.3;SystemConstruction;34
8.4;Verification;34
8.5;Conclusion;36
8.6;References;36
9;An Improved Object Detection and Contour Tracking Algorithm Based on Local Curvature;37
9.1;Introduction;37
9.2;Active Contour Model;38
9.3;Proposed Algorithm;39
9.3.1;Curvature Threshold and Inserting Additional Snake Points;39
9.3.2;Object Contour Tracking;40
9.3.3;Inserting and Deleting Snake Points;40
9.4;Experiments;41
9.4.1;Performance of Detection Analysis and Comparison;41
9.4.2;Performance of Tracking Analysis and Comparison;42
9.5;Conclusions;44
9.6;References;44
10;An Efficient Method for Noisy Cell Image Segmentation Using Generalized a-Entropy;45
10.1;Introduction;45
10.2;Entropy of Generalized Distributions;46
10.3;Proposed Segmentation Method;48
10.3.1;Preprocessing;48
10.3.2;Entropies Calculation;48
10.3.3;Image Thresholding;48
10.3.4;Morphology-Based Operations;49
10.3.5;Overlapping Cancelation;49
10.3.6;Non-objects Removal;49
10.4;Experimental Results;49
10.5;Conclusion;51
10.6;References;52
11;An Algorithm for Moving Multi-target Prediction in a Celestial Background;53
11.1;Introduction;53
11.2;Prediction of Moving Multi-targets;54
11.2.1;Targets Capture;54
11.2.2;Calculation of Targets Number and Coordinates;54
11.2.3;Image Block Processing;55
11.2.4;Target Prediction;56
11.3;Algorithm Simulation;57
11.4;Conclusion;58
11.5;References;59
12;Automatic Control Signal Generation of a CCD Camera for Object Tracking;60
12.1;Introduction;60
12.2;Signal to Control CCD Cameras;61
12.3;Creation of CCD Camera Control Signals;61
12.3.1;Creation of Pan Signals;62
12.3.2;Creation of Tilt Signals;63
12.4;Object Tracking Test of the CCD Camera;63
12.4.1;Initial Starting and Acceleration of CCD Camera Pan/Tilt;64
12.4.2;Effect of Frame Frequency;64
12.4.3;Effect of the Size of Pan/Tilt Signals;65
12.4.4;Object Tracking of CCD Camera;66
12.5;Conclusion;67
12.6;References;67
13;An Efficient Approach for Region-Based Image Classification and Retrieval;68
13.1;Introduction;68
13.2;Multi-level Neural Networks;69
13.3;Image Segmentation;70
13.4;Feature Extraction;71
13.4.1;Wavelet Decomposition;71
13.4.2;Color Moments;71
13.5;Proposed Approach;72
13.5.1;Preprocessing;72
13.5.2;Segmentation;73
13.5.3;Feature Extraction;73
13.5.4;Feature Normalization;73
13.5.5;Classification of Image Regions;73
13.6;Experimental Results;74
13.7;Conclusion;75
13.8;References;75
14;An Improved Design for Digital Audio Effect of Flanging;77
14.1;Introduction;77
14.2;Generation of Flanging Effect;77
14.2.1;The General Model;78
14.2.2;The Modified Model;78
14.3;Simulation Experiments and Result Analysis;79
14.3.1;Comparison of Modified Model with General Model;79
14.3.2;Performances of Modified Model for Different Modulation Waveforms;81
14.4;Conclusion;82
14.5;References;82
15;Robust Speech Enhancement Using Two-Stage Filtered Minima Controlled Recursive Averaging;84
15.1;Introduction;84
15.2;Overview of the Filtering Stage;85
15.2.1;Concept of Pre-processing;85
15.2.2;Spectral Gain Calculation;87
15.3;Enhancement Using FMCRA;87
15.3.1;Validation of the Two-Stage Progress;88
15.3.2;Noise Estimation;88
15.4;Experiments;89
15.5;Conclusions;91
15.6;References;92
16;Automatic Colon Cleansing in CTC Image Using Gradient Magnitude and Similarity Measure;94
16.1;Introduction;94
16.2;Proposed Method;95
16.2.1;Preprocessing;95
16.2.2;Histogram Peak Classification;96
16.2.3;Similarity Measure Classification;97
16.2.4;AT vs. ATT Layers;98
16.2.5;Reconstruction;99
16.3;Experimental Results;99
16.4;Discussion and Conclusions;100
16.5;References;101
17;A Scale and Rotation Invariant Interest Points Detector Based on Gabor Filters;102
17.1;Introduction;102
17.2;Gabor Functions;103
17.3;Rotation Invariant Interest Points Detection;104
17.4;Scale Selection in Gabor Scale-Space;105
17.5;Scale Invariant Interest Points Detection;106
17.6;Comparative Evaluation;107
17.7;Conclusion and Future Work;108
17.8;References;109
18;Signature Verification Based on Handwritten Text Recognition;110
18.1;Introduction;110
18.1.1;Related Work;111
18.2;Signature Recognition System;111
18.2.1;Preprocessing;111
18.2.2;Feature Extraction;111
18.2.3;Feature Normalization;115
18.2.4;Signature Verification;115
18.3;Experimental Results and Discussions;116
18.4;Conclusions;117
18.5;References;117
19;Effect of Image Linearization on Normalized Compression Distance;118
19.1;Introduction;118
19.2;Methodology;120
19.2.1;Kolmogorov Complexity and Derivations;120
19.2.2;Linearization Methods;121
19.2.3;Null Hypothesis Test;122
19.2.4;Dataset and Transformations;123
19.2.5;CompLearn;124
19.2.6;Statistical Analysis;125
19.3;Results;125
19.4;Discussion and Conclusion;126
19.4.1;Future Work;127
19.5;References;128
20;A Comparative Study of Blind Speech Separation Using Subspace Methods and Higher Order Statistics;129
20.1;Introduction;129
20.2;ICA Model;130
20.3;Blind Speech Separation Using High-Order Statistics;131
20.4;Separation of Mixed Speech Signals by OPCA;132
20.5;Experiments and Results;133
20.6;Conclusion;135
20.7;References;135
21;Automatic Fatigue Detection of Drivers through Yawning Analysis;137
21.1;Introduction;137
21.2;Literature Review;138
21.3;Proposed Approach;139
21.3.1;Image Acquisition;139
21.3.2;Face Detection and Tracking;139
21.3.3;Mouth Detection and Pupil Detection;140
21.3.4;Yawning Analysis and Driver Fatigue Monitoring;143
21.4;Experimental Results;143
21.5;Conclusion and Future Work;143
21.6;References;144
22;GA-SVM Based Lungs Nodule Detection and Classification;145
22.1;Introduction;145
22.2;Related Work;146
22.3;Proposed Method;147
22.3.1;Threshold Block;147
22.3.2;Background Removal;148
22.3.3;Edge Detection and Thinning;148
22.3.4;Lungs Border Identification and Reconstruction;148
22.3.5;Filling and Lungs Part Extraction;148
22.3.6;Histogram Based Threshold and Cleaning;148
22.3.7;Extraction of Region of Interests (ROIs);149
22.3.8;Pruning of ROIs;150
22.3.9;Features Extraction and Classification;150
22.4;Experimental Results and Discussion;150
22.5;References;152
23;Link Shifting Based Pyramid Segmentation for Elongated Regions;153
23.1;Introduction;153
23.2;Literature Review;155
23.2.1;Overview;155
23.2.2;Segmentation Algorithms;156
23.2.3;Region Similarity and Unforced Linking;156
23.3;Pyramid Image Segmentation Algorithm;157
23.3.1;Creating the Initial Image Pyramid;157
23.3.2;Testing Similarity of Two Regions, Unforced Linking and Tie-Breaking Rule;157
23.3.3;Candidate Parents;158
23.3.4;Pyramid Segmentation;159
23.4;Experimental Results;160
23.4.1;Segmentation of Basic Shapes;160
23.4.2;Segmentation of Other Imagery;161
23.5;Future Work;163
23.6;References;163
24;A Robust Algorithm for Fruit-Sorting under Variable Illumination;165
24.1;Introduction;165
24.2;Experimental Set-Up;166
24.3;Methodology;166
24.3.1;Pre-processing;167
24.3.2;Defect Segmentation;168
24.4;Results;169
24.5;Conclusion;171
24.6;References;171
25;Selection of Accurate and Robust Classification Model for Binary Classification Problems;173
25.1;Introduction;173
25.2;Literature Review;174
25.2.1;One Class Classifier;174
25.2.2;Two Class Classifier;175
25.3;Experiments;175
25.3.1;Experimental Setup;175
25.3.2;Data Sets;175
25.3.3;Multi Class Classifier;175
25.3.4;One Class Classifier;176
25.4;Results and Discussion;176
25.4.1;Receiver Operating Curve (ROC);177
25.4.2;Statistics (Pair Wise Measure);177
25.4.3;Cross Validation Error;179
25.5;Conclusion;179
25.6;References;179
26;Robust Edge-Enhanced Fragment Based Normalized Correlation Tracking in Cluttered and Occluded Imagery;181
26.1;Introduction;181
26.2;EEFNC Tracking Framework;182
26.3;Experimental Results;185
26.4;Conclusion;187
26.5;References;188
27;Robust and Imperceptible Watermarking of Video Streams for Low Power Devices;189
27.1;Introduction;189
27.2;Background;190
27.3;Proposed Method;192
27.4;Test Results;193
27.5;Conclusions and Future Work;194
27.6;References;196
28;A New Ensemble Scheme for Predicting Human Proteins Subcellular Locations;197
28.1;Introduction;197
28.2;Materials and Methods;198
28.2.1;Proposed Method;198
28.2.2;Evaluation Methods;200
28.3;Results and Discussions;201
28.4;Conclusion;202
28.5;References;203
29;Designing Linear Phase FIR Filters with Particle Swarm Optimization and Harmony Search;205
29.1;Introduction;205
29.2;Particle Swarm Optimization;207
29.3;Harmony Search Optimization;208
29.4;Case Studies and Discussion;209
29.5;Conclusion;211
29.6;References;212
30;Blind Image Steganalysis Based on Local Information and Human Visual System;213
30.1;Introduction;213
30.2;Segmentation and Clustering the Segments;214
30.3;Feature Extraction;216
30.4;Classification;217
30.5;Experiment Result;217
30.6;Conclusions;219
30.7;References;219
31;An Effective Combination of MPP Contour-Based Features for Off-Line Text-Independent Arabic Writer Identification;221
31.1;Introduction;221
31.2;Proposed Approach;223
31.3;Feature Extraction;223
31.3.1;Weighted Edge Direction Probability Distribution Function $(f1)$;225
31.3.2;Edge Length/Direction Probability Distribution Function $(f2)$;226
31.3.3;Angle Probability Distribution Function $(f3)$;226
31.3.4;Angle Co-occurrence Probability Distribution Function $(f4)$;226
31.3.5;Cross-Correlation of Angle Co-occurrence Distribution $(f5)$;227
31.3.6;Curvature Probability Distribution Function $(f6)$;228
31.4;Experimental Results;228
31.5;Conclusion;230
31.6;References;231
32;Ringing Artifact Removal in Digital Restored Images Using Multi-Resolution Edge Map;233
32.1;Introduction;233
32.2;Ringing Artifacts Removal Using the DWT-Based Adaptive Edge Map;234
32.2.1;Wavelet Analysis for Extracting Edge Region;234
32.2.2;Ringing Artifacts Removal;235
32.3;Experimental Results;237
32.4;Conclusion;238
32.5;References;239
33;Upmixing Stereo Audio into 5.1 Channel Audio for Improving Audio Realism;240
33.1;Introduction;240
33.2;Audio Upmixing Algorithm from Stereo to 5.1 Channel Audio;241
33.2.1;Passive Surround Decoding Method;241
33.2.2;LMS-Based Upmixing Method;242
33.2.3;PCA-Based Upmixing Method;242
33.2.4;Adaptive Panning Method;243
33.2.5;Low-Pass Filters;244
33.2.6;Time Delay and ±90° Phase Shifter;244
33.3;Design and Implementation of Audio Upmixing Simulator;245
33.4;Performance Evaluation;245
33.5;Conclusion;246
33.6;References;247
34;Multiway Filtering Based on Multilinear Algebra Tools;248
34.1;Introduction;248
34.2;Tensor Representation and Properties;250
34.3;Tensor Filtering Problem Formulation;251
34.3.1;Channel-by-Channel SVD-Based Filtering;252
34.3.2;Tensor Filtering Based on Multimode PCA;253
34.3.3;Multiway Wiener Filtering;255
34.4;Simulation Results;257
34.4.1;Performance Criterion;258
34.4.2;Denoising of Color Images;258
34.5;Conclusion;259
34.6;References;260
35;LaMOC – A Location Aware Mobile Cooperative System;262
35.1;Introduction;262
35.2;Scenario Based Design and Development of LaMOC;263
35.3;LaMOC System;265
35.3.1;Architecture of LaMOC;265
35.3.2;Map Based Browser and LaMOC User Interface;266
35.3.3;Fixed Host Layer in LaMOC;267
35.4;Some Key Issues;268
35.4.1;Context Awareness;268
35.4.2;Mobile Cooperation;269
35.5;Conclusion;270
35.6;References;270
36;Modelling of Camera Phone Capture Channel for JPEG Colour Barcode Images;271
36.1;Introduction;271
36.1.1;Errors Pertinent to JPEG Images Captured on Camera Phones;272
36.2;Camera Phone Capture Channel Modelling;273
36.3;Simulation Results;275
36.4;Conclusion;276
36.5;References;278
37;Semantic Network Adaptation Based on QoS Pattern Recognition for Multimedia Streams;279
37.1;Introduction;279
37.2;Multimedia Network Context: RTP-Based Systems;280
37.2.1;Fixed RTP Header;280
37.2.2;Specific RTP Profiles;281
37.3;Implicit Packet Meta Header Ontology;282
37.3.1;IPMH Structure;283
37.3.2;Mapping Rules;283
37.3.3;Using the IPMH;284
37.4;Case Study;285
37.5;Conclusions;286
37.6;References;286
38;Active Shape Model-Based Gait Recognition Using Infrared Images;287
38.1;Introduction;287
38.2;Active Shape Model for Object Extraction in Infrared Images;288
38.2.1;Shape Variation Modeling;288
38.2.2;Model Fitting;288
38.2.3;Local Structure Modeling;289
38.3;Extraction of Gait Data;289
38.4;Experiment Results;290
38.5;Conclusion;293
38.6;References;293
39;About Classification Methods Based on Tensor Modelling for Hyperspectral Images;294
39.1;Introduction;294
39.2;Matrix Algebra-Based DR Method;295
39.2.1;HSI Representation;295
39.2.2;Principal Component Analysis Based DR Approach;296
39.2.3;Independent Component Analysis Based DR Approach;296
39.2.4;Projection Pursuit Based DR Approach;297
39.3;Tensor Representation and Some Properties;297
39.4;Multilinear Algebra-Based DR Method;298
39.4.1;Tensor Formulation of $PCA_{dr}$ and $PP_{dr}$;298
39.4.2;Multilinear Algebra and $PCA$-Based DR Method;300
39.4.3;Multilinear Algebra and $PP$-Based DR Method;301
39.5;Experimental Results;302
39.5.1;Experiment on Simulated Data;302
39.5.2;Experiment on Real-World Data;303
39.6;Conclusion;306
39.7;References;307
40;Comparative Analysis of Wavelet-Based Scale-Invariant Feature Extraction Using Different Wavelet Bases;309
40.1;Introduction;309
40.2;Comparison of Feature Extraction Performance Using Different Wavelet Bases;310
40.2.1;Feature Extraction Using the Daubechies, Haar Wavelets;310
40.2.2;Feature Extraction Using Gabor Wavelet;311
40.3;Experiment Results;312
40.4;Discussion;313
40.5;References;315
41;A Novel Video Segmentation Algorithm with Shadow Cancellation and Adaptive Threshold Techniques;316
41.1;Introduction;316
41.2;Baseline Mode;317
41.3;Shadow Cancellation Mode;319
41.4;Adaptive Threshold Mode;320
41.5;Experimental Results;321
41.6;Conclusion;323
41.7;References;323
42;Considerations of Image Compression Scheme Hiding a Part of Coded Data into Own Image Coded Data;324
42.1;Introduction;324
42.2;Basic Principles of the Proposed Data Hiding Scheme;325
42.3;Image Coding Methods Using the Proposed Scheme;326
42.3.1;Fractal Image Coding Method;326
42.3.2;Image Coding Using Vector Quantization;327
42.3.3;Inter-frame Coding Using Motion Compensation;329
42.4;Considerations;330
42.5;Conclusion;330
42.6;References;331
43;Binarising SIFT-Descriptors to Reduce the Curse of Dimensionality in Histogram-Based Object Recognition;332
43.1;Introduction;332
43.2;Distance between SIFT-Descriptors;332
43.3;Proposed Method;335
43.4;Experimental Validation;336
43.5;Conclusion;338
43.6;References;339
44;Author Index;340

Leseproben

"A Robust Algorithm for Fruit-Sorting under Variable Illumination thresholding. (p. 153-154)

Abstract. Computer vision techniques are essential for defect segmentation in any automatized fruit-sorting system. Conventional sorting methods employ algorithms that are specific to standard illumination conditions and may produce undesirable results if ideal conditions are not maintained. This paper outlines a scheme that employs adaptive filters for pre-processing to negate the effect of varying illumination followed by defect segmentation using a localized adaptive threshold in an apple sorting experimental system based on a reference image. This technique has been compared with other methods and the results indicate an improved sorting performance. This can also be applied to other fruits with curved contours.

Keywords: Computer Vision, On-line fruit sorting, Surface defect, Adaptive thresholding.

1 Introduction

Fruit inspection and grading is an indispensable horticultural procedure. Uniformity in size, shape and colour are few of the many parameters that are vital in determining consumer acceptance. While the task at hand is to develop a machine vision system that identifies defective fruits based on odd shapes and surface defects, and to categorize them depending on consumer acceptability, the objective has to be accomplished with certain constraints [1].

Such a system has to be operable at high speeds suitable for real-time processing yielding a high throughput, must inspect the entire fruit surface, must be adaptable to varying fruit size, shape etc., and be applicable under various physical conditions like brightness, luminance etc. Over the past decade, various techniques have been proposed for defect segmentation and grading. Reference [2] uses flooding algorithm to identify and characterize different types of defects based on perimeter, area etc.

The snake algorithm discussed in [3] can be used to localize the defect and reduce false positives. Reference [4] employs a raw approach based on colour frequency distribution to associate pixels to a specific class while [5] accomplishes the same using ‘Linear discriminant analysis’ Hyper-spectral and multispectral imaging systems have also been proposed for sorting various food commodities as discussed in [6]-[7]. An inherent limitation in most of the existing techniques is their sensitivity to changing illumination conditions.

Any flash of external stimulus can result in bright patches in the captured image which could result in misclassification. Practical considerations dictate that any technique should be immune to occasional changes in external conditions and deliver acceptable performance. This paper incorporates the use of adaptive filters based on the conventional LMS algorithm as a pre-processing step prior to segmenting defects using an adaptive threshold. This paper has been organized as follows. Section 2 explains the components of the practical set-up used to capture images of the fruit to be sorted. Section 3 discusses the proposed methodology for pre-processing and defect segmentation. Results of the experiment have been tabulated and discussed towards the end."