About

The course is based on mathematical methods of pattern recognition used for the analysis and classification of images in computer vision systems. Image transformations are considered in order to generate feature descriptions. The methods of point, spatial geometric, algebraic and inter-frame image processing are studied. Methods of feature generation based on the decomposition of images by basic functions, statistical analysis of image texture, as well as image shape analysis are considered. Methods of construction of metrics for image comparison (comparison of spectral decompositions, overlay and alignment of images) are considered. And also questions of application of the studied methods in applied problems of computer vision. The tasks of text recognition in document images, the tasks of biometric identification of an individual by the texture of the iris, by the shape of the palm, fingerprint, face profile are considered.

Syllabus

• Subject and tasks of digital image processing and recognition.
• Spatial image processing methods.
• Geometric and algebraic methods of image processing.
• Methods of inter-frame image processing.
• Image analysis based on decomposition by basic functions.
• Statistical methods of texture analysis.
• Methods of image shape analysis.
• Metrics for measuring image similarity.
• Text recognition based on document images.
• Biometric identification based on image recognition.
• Dynamic scene recognition.

Labworks

Homeworks on image processing and classification.

Grading

The assessment is based on the results of laboratory work and an oral exam.

Prerequisites

Machine learning, linear algebra.