LEARNING OUTCOMES

The course aims at providing knowledge on theory and tools on the basics of Computer Vision, for the extraction of semantic and geometric information about a scene from an image or a sequence of images. Topics of interest include: camera models and image formation; camera calibration; connection between 2D images and 3D scene structures; image processing basics as image filtering, local features extraction (edge, corner, blob), including the use of multi-scale image representations; image matching, with reference to classification and retrieval problems; stereo vision and scene depth estimation; motion detection in image sequences, including change detection and optical flow estimation.

AIMS AND LEARNING OUTCOMES

The aim of the course is to provide a broad introduction to different core aspects of computer vision, including  camera modelling, camera calibration, image processing, pose estimation, multi view geometry, visual tracking, and vision based calibration.

At the end of the course the student will be able to understand the main theoretical concepts and to design and implement classical computer vision algorithms. The course will also provide an overview of the main application domains, with a special reference to the robotics scenario.

PREREQUISITES

background knowledge on linear algebra and calculus; basic programming skills

TEACHING METHODS

• theoretical classes followed by practical (step-by-step) activities that will be online

SYLLABUS/CONTENT

Introduction to computer vision for robotics applications

Part 1 - image processing fundamentals

• Digital image fundamentals: sensing and acquisition, sampling and quantization, basic operations (warping)
• Intensity transformations and spatial filtering (filtering in the frequency domain) 
• Edge and corner detection
• Color image processing
• Hough transforms and image segmentation
• Scale space and blob detection
• Image matching

Part 2 - motion analysis 

• Motion: 3D and 2D motion fields, dense and sparse optical flow. Dominant motion estimation                              
• Tracking with linear dynamic models (Kalman Filter)

Part 3 - geometry

• 3D computer vision fundamentals
• The geometry of image formation: review of projective geometry (basic), projective transformations, camera models and single view geometry, camera calibration, Homographies
• Stereopsis: epipolar geometry, stereo rectification, depth estimation, 3D reconstruction 

​Conclusions: Visual Recognition and  image retrieval;  introduction to object and action recognition methods in HRI

RECOMMENDED READING/BIBLIOGRAPHY

Recommended texts:

• R.C. Gonzalez and R.E. Woods, Digital image processing, Prentice-Hall, 2008.
• E. Trucco and A. Verri, Introductory Techniques for 3-D Computer Vision, Prentice Hall,  1998.

Further readings: Material distributed by lecturers through the Aulaweb portal