Скачать презентацию Technion — Israel Institute of Technology Department of Скачать презентацию Technion — Israel Institute of Technology Department of

8e9ed197d577ac63a822e44627c0ad91.ppt

  • Количество слайдов: 62

Technion - Israel Institute of Technology Department of Electrical Engineering Vision & Image Sciences Technion - Israel Institute of Technology Department of Electrical Engineering Vision & Image Sciences Laboratory http: // visl. technion. ac. il November 2003

Introduction Fields of interest Vision Research Computer Vision Image Processing Pattern Recognition Computer Graphics Introduction Fields of interest Vision Research Computer Vision Image Processing Pattern Recognition Computer Graphics Biological Signals

Introduction Academic Staff Prof. Y. Y. Zeevi Prof. R. Meir Dr. M. Porat Dr. Introduction Academic Staff Prof. Y. Y. Zeevi Prof. R. Meir Dr. M. Porat Dr. A. Tal Dr. M. Zibulevsky Dr. Y. Shechner Dr. Y. Eldar

Introduction Technical Staff Eng. Johanan Erez Eng. Eli Appelboim Eng. Ina Krinsky Tech. Aharon Introduction Technical Staff Eng. Johanan Erez Eng. Eli Appelboim Eng. Ina Krinsky Tech. Aharon Yacoby Student assistents

Introduction Activity • 10 -15 grad. students. • ~ 30 undergr. projects / semester. Introduction Activity • 10 -15 grad. students. • ~ 30 undergr. projects / semester. • ~100 lab experiments / semester.

Blind Source Separation Blind Source Separation

Blind Source Separation The blind source separation problem is to extract the underlying source Blind Source Separation The blind source separation problem is to extract the underlying source signals from a set of linear mixtures, where the mixing matrix is unknown. This situation is common in acoustics, radio, medical signal and image processing, hyperspectral imaging, etc.

 הפרדה )עיוורת( של תמונות Blind Source Separation MIX Signal sources BSS Mixed signals הפרדה )עיוורת( של תמונות Blind Source Separation MIX Signal sources BSS Mixed signals Separated signals

Blind Source Separation Applications: • Audio Signals Separation. • Mixed Images. • f. MRI Blind Source Separation Applications: • Audio Signals Separation. • Mixed Images. • f. MRI Images. • Biological Signals. • Hyperspectral Images. • …

Blind Separation of Audio Sources Audio Signals: § Sources § Mixed Signals § Separated Blind Separation of Audio Sources Audio Signals: § Sources § Mixed Signals § Separated Signals

Blind Separation of Images Blind Separation of Images

Blind Source Separation of f. MRI Images Blind Source Separation of f. MRI Images

Simulator f. MRI Simulator f. MRI

f. MRI Simulator - Results f. MRI Simulator - Results

BSS of f. MRI – Simulation Results BSS of f. MRI – Simulation Results

Blind Source Separation of f. MRI Images Blind Source Separation of f. MRI Images

Blind Source Separation of f. MRI Images Blind Source Separation of f. MRI Images

Blind Source Separation of EEG Signals Blind Source Separation of EEG Signals

Hyperspectral Analysis using Blind Source Separation Hyperspectral Analysis using Blind Source Separation

Combining classical Blind Source Separation with spatial approach to the Hyperspectral data. Combining classical Blind Source Separation with spatial approach to the Hyperspectral data.

Computer Vision Computer Vision

Recognition and Tracking of corners with SUSAN Algorithm Recognition and Tracking of corners with SUSAN Algorithm

Vision-based Door Control No target Closed door Opened door Approaching target Passing target Vision-based Door Control No target Closed door Opened door Approaching target Passing target

Slides from Video lecture Slides from Video lecture

Computer Vision Computer Vision

Computer Vision short spur dirt open mousebite pinhole Computer Vision short spur dirt open mousebite pinhole

LEGO Lab Wireless connection between vehicle and PC Vision based Navigating LEGO Lab Wireless connection between vehicle and PC Vision based Navigating

Autonomous Vehicles DEMO 1 DEMO 2 Autonomous Vehicles DEMO 1 DEMO 2

Autonomous Vehicles Autonomous Vehicles

Autonomous Vehicles Autonomous Vehicles

Soccer Game Tracking DEMO 1 DEMO 2 Soccer Game Tracking DEMO 1 DEMO 2

 גוף מתחת לקו חוף )או קו אופק(: גוף מתחת לקו חוף )או קו אופק(:

 סיכום סוגי התרחישים: ים מלא ים ומטרה קו חוף מטרה מתחת מטרה מעל סיכום סוגי התרחישים: ים מלא ים ומטרה קו חוף מטרה מתחת מטרה מעל

Naval Targets § Demo 1 § Demo 2 § Demo 3 Naval Targets § Demo 1 § Demo 2 § Demo 3

Pattern Recognition Pattern Recognition

Traffic Sign Recognition Traffic Sign Recognition

 הדגמה לסיווג הדגמה לסיווג

Classification by Support Vector Machines Classification by Support Vector Machines

 אחוזי סיווג לשלב תמרור בתוך קבוצה % 35. 29 % 76. 79 ? אחוזי סיווג לשלב תמרור בתוך קבוצה % 35. 29 % 76. 79 ? % 76. 79 % 4858. 89 ? Yes % 4858. 89 ? % 4858. 89 Road ? Sign % 4858. 89 % 7063. 89 ? No % 7063. 89 % 4858. 89 % 5008. 59 % 4858. 89 ?

License Plate Recognition License Plate Recognition

Neural network architecture 20 pixels 10 pixels Input Layer 20 x 10 = 200 Neural network architecture 20 pixels 10 pixels Input Layer 20 x 10 = 200 neurons Middle Layer 20 neurons Output Layer 10 neurons

Digital Cameras Adaptive Gain Control by i-sight Camera Digital Cameras Adaptive Gain Control by i-sight Camera

Adaptive Sensitivity Adaptive Sensitivity

Hardware Implementations using Trimedia Multimedia Processor Video IN Video OUT Hardware Implementations using Trimedia Multimedia Processor Video IN Video OUT

Superresolution Superresolution

Image Enhancement by Superresolution Source image מקור Enhanced image Image Enhancement by Superresolution Source image מקור Enhanced image

Superresolution - results Superresolution - results

Image Enhancement by Diffusion Image Enhancement by Diffusion

Indexing of image databases Indexing of image databases

Image Indexing by contour examination Which fish looks like: ? ? ? Image Indexing by contour examination Which fish looks like: ? ? ?

Contour Signature § The goal: to get a one-dimensional description of the image - Contour Signature § The goal: to get a one-dimensional description of the image - Signature. § The method: sampling points on the curve, and finding the outer angel of each three points. Sl(m) = 30° m m+1 m-1 k+1 k Sl(k)= -80°

Image Indexing - Results C) D) A) Image Indexing - Results C) D) A)

Indexing of images according to color signature Indexing of images according to color signature

Image Indexing Image Indexing