Humans use vision as a primary source to obtain information about the outside world. The goal of computer vision is to implement similar functionalities in machines by developing algorithms and computational models to automatically process and extract information from digital image data. This course offers an introduction to the field that has grown considerably in recent years, much due to the decreased cost of digital photography and the increasing amount of visual information on Internet and elsewhere. Application areas include e.g. robot vision, medical imaging, automated inspection, three-dimensional modeling, human-computer interaction, image compression and interpretation of aerial and satellite images.
In the course, you will initially learn basic image operations to enhance and extract information from digital images. Examples are gray-level transformations, filtering techniques and detection of features such as corners, edges and regions (segmentation). We will also study methods to derive three-dimensional information about the outside world on the basis of visual information, using cues such as texture, shading, stereo and motion. In terms of robotics, we will e.g. look at how to make a robot avoid collisions by just computing a simple descriptor directly from raw image data. We will further study methods for object recognition and talk about their complexities and inherent limitations. Computer vision is a strongly inter-disciplinary subject with strong connections to theories on biological and human vision, as well as to machine learning and other related areas.