computer vision reu week 2 adam kavanaugh. video canny put canny into a loop in order to process...

14
Computer Vision REU Computer Vision REU Week 2 Week 2 Adam Kavanaugh Adam Kavanaugh

Upload: marjorie-page

Post on 18-Dec-2015

222 views

Category:

Documents


0 download

TRANSCRIPT

Computer Vision REUComputer Vision REUWeek 2Week 2

Adam KavanaughAdam Kavanaugh

Video CannyVideo Canny

Put canny into a loop in order to process Put canny into a loop in order to process multiple frames of a video sequencemultiple frames of a video sequence

Output each individual picture to its own Output each individual picture to its own directory for ease of processingdirectory for ease of processing

10 Research Topics10 Research Topics

Search the Vision group sites of:Search the Vision group sites of:Cornell UniversityCornell UniversityCarnegie Mellon UniversityCarnegie Mellon UniversityStanford UniversityStanford UniversityUniversity of CambridgeUniversity of Cambridge

Cornell UniversityCornell University

Detection and long term tracking of Detection and long term tracking of moving objects in aerial videomoving objects in aerial video

Cornell UniversityCornell University

Model based tracking in quasi real-timeModel based tracking in quasi real-time

Cornell UniversityCornell University

Recognizing Flexible Objects Recognizing Flexible Objects

Carnegie Mellon UniversityCarnegie Mellon University

AAM Fitting AlgorithmsAAM Fitting AlgorithmsActive Appearance ModelsActive Appearance Models (AAM) (AAM)

Carnegie Mellon UniversityCarnegie Mellon University

Gaze EstimationGaze Estimation Uses an AAM algorithm to detect features like Uses an AAM algorithm to detect features like

eye cornerseye corners

Carnegie Mellon UniversityCarnegie Mellon University

2D->3D Face Model Construction2D->3D Face Model Construction Uniquely recovers the 3D non-rigid shapes Uniquely recovers the 3D non-rigid shapes

and poses of a human face from a 2D and poses of a human face from a 2D monocular video monocular video

Carnegie Mellon UniversityCarnegie Mellon University

Scene FlowScene Flow Scene flow is the three-dimensional motion Scene flow is the three-dimensional motion

field of points in the worldfield of points in the worldOptical flow in 3D! Optical flow in 3D!

Stanford UniversityStanford University

Edge, Corner, and Junction Detection with Edge, Corner, and Junction Detection with the Generalized Compass Operator the Generalized Compass Operator Find edges were Canny failsFind edges were Canny fails

Stanford UniversityStanford University

Elliptical Head Tracking Using Intensity Elliptical Head Tracking Using Intensity Gradients and Color HistogramsGradients and Color Histograms In real time, the tracker is able to reliably and In real time, the tracker is able to reliably and

automatically control the camera's pan, tilt, automatically control the camera's pan, tilt, and zoom in order to keep the subject and zoom in order to keep the subject centered in the field of view at a desired size. centered in the field of view at a desired size.

University of CambridgeUniversity of Cambridge

Image Divergence  from Closed CurvesImage Divergence  from Closed Curves Finds time to contact and surface orientation Finds time to contact and surface orientation

reliablyreliablyThis is exploited in real-time visual docking This is exploited in real-time visual docking

and obstacle avoidance. and obstacle avoidance.

Lucas Kanade C conversionLucas Kanade C conversion

In Progress…In Progress…On line 71 of the MATLAB fileOn line 71 of the MATLAB fileCompiles but no runtime tests have been Compiles but no runtime tests have been

mademadeC file is already twice as large as the C file is already twice as large as the

MATLAB fileMATLAB file I understand why Alex recommended I understand why Alex recommended

doing this coding in MATLAB!doing this coding in MATLAB!