Vision Guided Control (Applied Early Vision)

COMPSCI 773 S1 T

Introduction

This course introduces computational methods and techniques used in popular vision-based research areas such as 2/3D face recognition, 3D scene reconstruction, 2D and 3D hand posture recognition that ar enow at the forefront of HCI and 3D Vision applications in the industry . Many topics are only reviewed, but a number of interesting theoretical and practical problems are analysed in detail. You will be able to acquire knowledge currently used in the latest technological advances available. This course is a must for students eager to pursue post-graduate studies and/or a career in applied computer vision.

Design of modern control systems involves different mathematical tools, especially, optimisation techniques, matrix analysis, and analytic 2D/3D geometry. Some tools will be explained in brief in the lectures. Still, you are expected to learn these methods in details and use them to complete assignments.

Programming in Computer Vision is usually undertaken in C, C++, C# . You are expected to be pro-efficient in at least one of the above mentioned programming language or show a strong willingness to learn.

In 2009, we are offering the opportunity to program Assignment 1 and tentatively Assignment 2 in Java using classes provided by the lecturers. It is expected that your programs will be uploaded and will have to work with the following Online Stereo Vision application  

Assessment

Assessment is based on 60% course work (30% group work, 30% individual work) and 40% open-book final examination. Course work includes one-to-one oral test  with the lecturers and assignments that exploit the digital cameras (Quickcam Pro 9000) available in our research labs at Tamaki (room 731.234). For each assignment, each group will have to write a report which should be organised as follows:

Course work

A particular feature of the course work is the emphasis on complete system design. Therefore, instead of picking a small part of the material covered in lectures as assignment tasks, the project in this paper has the aim of developing a complete system to perform a specified task. The individual assignments present intermediate steps toward achieving this goal. At the end of the paper, each group will present his completed project.

The course relies on the Postgraduate room for lab sessions and equipment, namely, a number of PCs running Windows and a few web-cameras to complete your project.  We may also use our 3D scanner and/or stereo-vision systems for 3D face acquisition.

This year course projects will encompass topics such as stereo vision, camera calibration, image pairs rectification, feature extraction and classification with a focus on depth maps and recognition tasks based on 2D and 3D (depth) image information. Basically, you will have to:

• Implement matching algorithms, display depth map, visualise the 3D data (optional) 
• Design an interface to acquire synchronised images from 2 USB cameras.
• Calibrate cameras, rectify image pairs
• Authenticate faces in the labs
• Extract faces from images and identify them using advanced statistical analysis techniques such as PCA, LDA.
• Fuse stereo-vision face data (depth map) and readily available face texture for 2+3D face recognition

The work is subdivided into three assignments (see Assignments panel) covering the following parts of the project:

1. Image / stereo matching algorithms.
2. Calibration of stereo cameras for computing 3D positions of a desired item in the cameras field-of-view(you will use in this assignment the existing Tsai calibration software (or any suitable calibration method you might have researched).
3. You will use statistical analysis techniques to recognise face.
1. Face authentication for faces localisation, face mask extraction, and face recognition at real-time.
2. 2+3D face authentication (fusing depth map and face texture images)
3. Whole system testing.

The schedule of these assignments is as follows:

Basic Topics of the Course

1. Low-level Image processing (for handouts: go to "Lectures")
   • Colour detection and discrimination
   • Binary image segmentation
   • Image matching
2. Stereo-vision 
   •  Stereo matching
3. Camera calibration and projective geometry (for handouts: go to "Lectures")
   • 2D and 3D vision geometry
   • Single camera calibration
   • Stereo calibration: epipolar geometry, traingulation, rectification
4. Face expression and face recognition
•  Feature extraction: face
•  Feature classification (PCA, LDA)
5. John Morris material

773 Groups