Signal Processing Institute Swiss Federal Institute of Technology, Lausanne 1 “OBJECTIVE AND SUBJECTIVE IDENTIFICATION OF INTERESTING AREAS IN VIDEO SEQUENCES”

Slides:



Advertisements
Similar presentations
Road-Sign Detection and Recognition Based on Support Vector Machines Saturnino, Sergio et al. Yunjia Man ECG 782 Dr. Brendan.
Advertisements

Evaluating Color Descriptors for Object and Scene Recognition Koen E.A. van de Sande, Student Member, IEEE, Theo Gevers, Member, IEEE, and Cees G.M. Snoek,
Change Detection C. Stauffer and W.E.L. Grimson, “Learning patterns of activity using real time tracking,” IEEE Trans. On PAMI, 22(8): , Aug 2000.
1 Human-Computer Interaction Screen Layout and Colour.
Contrast-Aware Halftoning Hua Li and David Mould April 22,
R&D Forum - 22 maggio 2009 Image Processing Laboratory DEEI, University of Trieste, Italy
Mahmoud Abdallah Daniel Eiland. The detection of traffic signals within a moving video is problematic due to issues caused by: Low-light, Day and Night.
Color spaces CIE - RGB space. HSV - space. CIE - XYZ space.
COLORCOLOR A SET OF CODES GENERATED BY THE BRAİN How do you quantify? How do you use?
AlgirdasBeinaravičius Gediminas Mazrimas Salman Mosslem.
Current Trends in Image Quality Perception Mason Macklem Simon Fraser University
ICME 2008 Huiying Liu, Shuqiang Jiang, Qingming Huang, Changsheng Xu.
Recognition of Traffic Lights in Live Video Streams on Mobile Devices
Qian Chen, Guangtao Zhai, Xiaokang Yang, and Wenjun Zhang ISCAS,2008.
Robust Object Segmentation Using Adaptive Thresholding Xiaxi Huang and Nikolaos V. Boulgouris International Conference on Image Processing 2007.
Face Detection: a Survey Speaker: Mine-Quan Jing National Chiao Tung University.
Introduction to Image Quality Assessment
Visual Attention More information in visual field than we can process at a given moment Solutions Shifts of Visual Attention related to eye movements Some.
Vision Computing An Introduction. Visual Perception Sight is our most impressive sense. It gives us, without conscious effort, detailed information about.
Highlights Lecture on the image part (10) Automatic Perception 16
Visual Information Retrieval Chapter 1 Introduction Alberto Del Bimbo Dipartimento di Sistemi e Informatica Universita di Firenze Firenze, Italy.
CSE 291 Final Project: Adaptive Multi-Spectral Differencing Andrew Cosand UCSD CVRR.
The Human Visual System Vonikakis Vasilios, Antonios Gasteratos Democritus University of Thrace
A Novel 2D To 3D Image Technique Based On Object- Oriented Conversion.
California Car License Plate Recognition System ZhengHui Hu Advisor: Dr. Kang.
Jacinto C. Nascimento, Member, IEEE, and Jorge S. Marques
Image Subtraction for Real Time Moving Object Extraction Shahbe Mat Desa, Qussay A. Salih, CGIV’04.
Automatic Estimation and Removal of Noise from a Single Image
Speaker: Chi-Yu Hsu Advisor: Prof. Jian-Jung Ding Leveraging Stereopsis for Saliency Analysis, CVPR 2012.
A Model of Saliency-Based Visual Attention for Rapid Scene Analysis Laurent Itti, Christof Koch, and Ernst Niebur IEEE PAMI, 1998.
A Generic Virtual Content Insertion System Based on Visual Attention Analysis H. Liu 1, 2, S. Jiang 1, Q. Huang 1, 2, C. Xu 2, 3 1 Institute of Computing.
Motion Object Segmentation, Recognition and Tracking Huiqiong Chen; Yun Zhang; Derek Rivait Faculty of Computer Science Dalhousie University.
Content-Based Image Retrieval
1 Webcam Mouse Using Face and Eye Tracking in Various Illumination Environments Yuan-Pin Lin et al. Proceedings of the 2005 IEEE Y.S. Lee.
03/05/03© 2003 University of Wisconsin Last Time Tone Reproduction If you don’t use perceptual info, some people call it contrast reduction.
National Taiwan A Road Sign Recognition System Based on a Dynamic Visual Model C. Y. Fang Department of Information and.
1 Research Question  Can a vision-based mobile robot  with limited computation and memory,  and rapidly varying camera positions,  operate autonomously.
by Mitchell D. Swanson, Bin Zhu, and Ahmed H. Tewfik
Region-Based Saliency Detection and Its Application in Object Recognition IEEE TRANSACTIONS ON CIRCUITS AND SYSTEM FOR VIDEO TECHNOLOGY, VOL. 24 NO. 5,
Image-Based Segmentation of Indoor Corridor Floors for a Mobile Robot Yinxiao Li and Stanley T. Birchfield The Holcombe Department of Electrical and Computer.
Introduction to Computer Graphics
Image Emotional Semantic Query Based On Color Semantic Description Wei-Ning Wang, Ying-Lin Yu Department of Electronic and Information Engineering, South.
Autonomous Robots Vision © Manfred Huber 2014.
Segmentation of Vehicles in Traffic Video Tun-Yu Chiang Wilson Lau.
Spatiotemporal Saliency Map of a Video Sequence in FPGA hardware David Boland Acknowledgements: Professor Peter Cheung Mr Yang Liu.
By Naveen kumar Badam. Contents INTRODUCTION ARCHITECTURE OF THE PROPOSED MODEL MODULES INVOLVED IN THE MODEL FUTURE WORKS CONCLUSION.
Multiple Light Source Optical Flow Multiple Light Source Optical Flow Robert J. Woodham ICCV’90.
Colour and Texture. Extract 3-D information Using Vision Extract 3-D information for performing certain tasks such as manipulation, navigation, and recognition.
Spatio-temporal saliency model to predict eye movements in video free viewing Gipsa-lab, Grenoble Département Images et Signal CNRS, UMR 5216 S. Marat,
Demosaicking for Multispectral Filter Array (MSFA)
Journal of Visual Communication and Image Representation
Suspicious Behavior in Outdoor Video Analysis - Challenges & Complexities Air Force Institute of Technology/ROME Air Force Research Lab Unclassified IED.
Optimal Eye Movement Strategies In Visual Search.
Learning video saliency from human gaze using candidate selection CVPR2013 Poster.
Chapter 24: Perception April 20, Introduction Emphasis on vision Feature extraction approach Model-based approach –S stimulus –W world –f,
Learning and Removing Cast Shadows through a Multidistribution Approach Nicolas Martel-Brisson, Andre Zaccarin IEEE TRANSACTIONS ON PATTERN ANALYSIS AND.
Image Perception ‘Let there be light! ‘. “Let there be light”
Color Models Light property Color models.
A. M. R. R. Bandara & L. Ranathunga
Visual Information Retrieval
Color Image Processing
Color Image Processing
DIGITAL SIGNAL PROCESSING
Color Image Processing
Error Concealment In The Pixel Domain And MATLAB commands
Enhanced-alignment Measure for Binary Foreground Map Evaluation
Davide Nardo, Valerio Santangelo, Emiliano Macaluso  Neuron 
Color Image Processing
Color Image Processing
Presentation transcript:

Signal Processing Institute Swiss Federal Institute of Technology, Lausanne 1 “OBJECTIVE AND SUBJECTIVE IDENTIFICATION OF INTERESTING AREAS IN VIDEO SEQUENCES” Danko Tomašić DIPLOMA PROJECT Erasmus Student, University of Trieste, Italy Responsible Assistant: Elisa Drelie Gelasca Professor: Touradj Ebrahimi

Signal Processing Institute Swiss Federal Institute of Technology, Lausanne 2 Outline  Introduction – Context – Application – State of the Art – Problem Statement  Subjective Experiments for Segmentation Quality Evaluation - Generation of Synthetic Segmentation Errors - Experimental Method - Data Analysis - Work in progress  Conclusions and Future Work

Signal Processing Institute Swiss Federal Institute of Technology, Lausanne 3 Motivation  REAL WORLD SCENES - presents a large quantity of information to the Human... - high visual acuity – fovea - eye movements – not random ! VISUAL ATTENTION Automatic method for.. REGIONS OF INTEREST (ROI) In video sequences SELECTIVITY

Signal Processing Institute Swiss Federal Institute of Technology, Lausanne 4 APPLICATIONS  COMPRESSION - preserve the quality in ROIs  SEGMENTATION QUALITY EVALUATION - overall and individual object  IMAGE AND VIDEO QUALITY ANALYSIS –Perceptual PSNR..  IMAGE AND VIDEO DATABASES - extract ROIs in each scene ‘a priori’  DIGITAL WATERMARKING - perceptual marking  MACHINE VISION - real-time robot navigation

Signal Processing Institute Swiss Federal Institute of Technology, Lausanne 5 FACTORS WHICH INFLUENCE VISUAL ATTENTION  low-level or bottom-up - motion - position - size - color and brightness - contrast - shape - orientation  high-level or top-down - presence of people - background / foreground rapid saliency driven task independent slower volition controlled Relevance.. task dependent

Signal Processing Institute Swiss Federal Institute of Technology, Lausanne 6 SALIENCE AND RELEVANCE  Salience - bottom-up task-independent factors - connected with observal-external objects or properties - unexpectedness or unusualness of an object  Relevance - top-down volition-controlled and task-dependent factors - observal-internal factors (goals and motivations) - connected to the specific situation = Conspicuity

Signal Processing Institute Swiss Federal Institute of Technology, Lausanne 7 PREVIOUS ATTENTION MODELS  Correia and Perreira - estimation of video object’s relevance - both individually and in a given context  Osberger - perceptual vision model for image quality assessment and compression applications - early vision model and higher level attention processes - Importance Maps

Signal Processing Institute Swiss Federal Institute of Technology, Lausanne 8 PREVIOUS MODELS  Osberger and Rohaly - automatic detection of ROIs in video sequences - improvement of previous model - adding temporal and movement features  Pardo - extraction of semantic objects from still images - perceptual metric and Importance Maps - improvement of Osberger’s models

Signal Processing Institute Swiss Federal Institute of Technology, Lausanne 9 COLOR – STATE OF THE ART  Osberger - red attracts attention more than other colors - red induces higher amount of masking  Correia and Perreira - bright and colored objects are more noticed - red seems to be preferred PROBLEM – NO OTHER QUANTIFIED SOLUTIONS !!!

Signal Processing Institute Swiss Federal Institute of Technology, Lausanne 10 COLOR – SIGNALLING AND MARKETING  American Standards Association and French Standard - red - dinamic, warm - fireproof protection - red - salient ‘par excellence’ - orange and yellow - salient colors – danger - green - calming color - materials of first aid - blue - cold and calming - caution PROBLEM – NO OTHER QUANTIFIED SOLUTIONS !!! “psychologically active”

Signal Processing Institute Swiss Federal Institute of Technology, Lausanne 11 Subjective Experiment on Color..  COLOR Color Test II Color Test III Color Test I

Signal Processing Institute Swiss Federal Institute of Technology, Lausanne 12 COLOR – COLOR TEST I First cycle Second cycle Tested colors in CIELab color space

Signal Processing Institute Swiss Federal Institute of Technology, Lausanne 13 COLOR TEST I - RESULTS

Signal Processing Institute Swiss Federal Institute of Technology, Lausanne 14 COLOR – COLOR TEST II Colored objects with semantical meaning Annoyance and Salience of a color interconnected ? Adding gaussian noise (sigma =...)to channel L in HSL color space Matlab - RANDN (M,N) - normal distribution mean = 0 variance = Rating the annoyance level

Signal Processing Institute Swiss Federal Institute of Technology, Lausanne 15 COLOR – COLOR TEST II Tested colors in CIELab color space Test image

Signal Processing Institute Swiss Federal Institute of Technology, Lausanne 16 COLOR TEST II - RESULTS

Signal Processing Institute Swiss Federal Institute of Technology, Lausanne 17 COLOR TEST II - RESULTS Cyan Green Pink Magenta Red Light blue Blue Orange Yellow = 241

Signal Processing Institute Swiss Federal Institute of Technology, Lausanne 18 COLOR – COLOR TEST III Colored objects with semantical meaning Annoyance and Salience of a color interconnected ? Blurred images - 3 levels of blurriness Rating the annoyance level

Signal Processing Institute Swiss Federal Institute of Technology, Lausanne 19 COLOR – COLOR TEST III Original imageTest image

Signal Processing Institute Swiss Federal Institute of Technology, Lausanne 20 COLOR TEST III - RESULTS COLOR TEST I vs. COLOR TEST III Second cycle highest level of blurriness 5 CASES Most important Most annoying 3 CASES 1st or 2nd most 2nd or 1st most important annoying 4 CASES RED 3rd most annoying !!! 4 CASES CYAN 1st most annoying !!!

Signal Processing Institute Swiss Federal Institute of Technology, Lausanne 21 COLOR TEST III - RESULTS COLOR TEST I vs. COLOR TEST III  Light blue  Violet  Dark Green  Maroon  Red  Yellow  Green  Blue Similar importance Similar annoyance

Signal Processing Institute Swiss Federal Institute of Technology, Lausanne 22 COLOR TEST III - RESULTS Average level lines 2 nd level of blurriness 3 rd level of blurriness

Signal Processing Institute Swiss Federal Institute of Technology, Lausanne 23 COLOR TEST III - RESULTS HSL color spaceCIELab color space

Signal Processing Institute Swiss Federal Institute of Technology, Lausanne 24 OBJECTIVE IDENTIFICATION OF INTERESTING AREAS Proposed method

Signal Processing Institute Swiss Federal Institute of Technology, Lausanne 25 SEGMENTATION  WATERSHED SEGMENTATION WATERSHED SEGMENTATION RESULT OF MERGE Sequence #1 - Akiyo Frame #15 ORIGINAL FRAME #15

Signal Processing Institute Swiss Federal Institute of Technology, Lausanne 26 FACTORS WHICH INFLUENCE ATTENTION  COLOR

Signal Processing Institute Swiss Federal Institute of Technology, Lausanne 27 FACTORS WHICH INFLUENCE ATTENTION  CONTRAST L, a, b – CIELab coordinates B i,j – border pixels in R i,j with R j k border = 10 R i,j – considered region R j – neighboring region

Signal Processing Institute Swiss Federal Institute of Technology, Lausanne 28 FACTORS WHICH INFLUENCE ATTENTION  SIZE

Signal Processing Institute Swiss Federal Institute of Technology, Lausanne 29 FACTORS WHICH INFLUENCE ATTENTION  POSITION

Signal Processing Institute Swiss Federal Institute of Technology, Lausanne 30 FACTORS WHICH INFLUENCE ATTENTION  SKIN HSV color space

Signal Processing Institute Swiss Federal Institute of Technology, Lausanne 31 FACTORS WHICH INFLUENCE ATTENTION  MOTION - OPTICAL FLOW Original video Movement mask

Signal Processing Institute Swiss Federal Institute of Technology, Lausanne 32 FACTORS WHICH INFLUENCE ATTENTION  FINAL COMBINATION OF FEATURES

Signal Processing Institute Swiss Federal Institute of Technology, Lausanne 33 SUBJECTIVE EXPERIMENTS  SEGMENTATION QUALITY DEFECT IN THE MOST IMPORTANT REGION DEFECT IN THE LEAST IMPORTANT REGION Sequence #1 - Akiyo Frame #15

Signal Processing Institute Swiss Federal Institute of Technology, Lausanne 34 SUBJECTIVE EXPERIMENTS  SEGMENTATION QUALITY DEFECT IN THE MOST IMPORTANT REGION DEFECT IN THE LEAST IMPORTANT REGION Sequence #2 - Highway Frame #28

Signal Processing Institute Swiss Federal Institute of Technology, Lausanne 35 SUBJECTIVE EXPERIMENTS  SEGMENTATION QUALITY DEFECT IN THE MOST IMPORTANT REGION DEFECT IN THE LEAST IMPORTANT REGION Sequence #3 - VideoEPFL Frame #39

Signal Processing Institute Swiss Federal Institute of Technology, Lausanne 36 SUBJECTIVE EXPERIMENTS  RESULTS Sequence #1 - Akiyo

Signal Processing Institute Swiss Federal Institute of Technology, Lausanne 37 SUBJECTIVE EXPERIMENTS  RESULTS Sequence #2 - Highway

Signal Processing Institute Swiss Federal Institute of Technology, Lausanne 38 SUBJECTIVE EXPERIMENTS  RESULTS Sequence #3 - VideoEPFL

Signal Processing Institute Swiss Federal Institute of Technology, Lausanne 39 VISUAL PERCEPTION - extracting information from the light  Acquiring knowledge - HVS = video camera  Objects and events in the environment  Light emitted or reflected by objects

Signal Processing Institute Swiss Federal Institute of Technology, Lausanne 40 VISUAL ATTENTION - processing different information within the visual field  Overt eye movements - determine available optic information  Covert selective attention - determines what gets full processing CAPACITY SELECTIVITY amount of available what gets processed perceptual resources and what does not

Signal Processing Institute Swiss Federal Institute of Technology, Lausanne 41 COLOR TEST I - RESULTS

Signal Processing Institute Swiss Federal Institute of Technology, Lausanne 42 COLOR TEST II - RESULTS Cyan Green Pink Magenta Red Light blue Blue Orange Yellow >>10

Signal Processing Institute Swiss Federal Institute of Technology, Lausanne 43 FACTORS WHICH INFLUENCE ATTENTION  COLOR Color Experiment 2 Color Experiment 3 Color Experiment 1