Course Overview

The topics covered in this course are of great importance and modernity regarding both biological vision as well as image processing and artificial vision. The integrated approach uses parallels between biological and computational systems, which is seldom covered in graduate courses in Brazil.

Learning Achievement

Familiarization with intermediate and advanced concepts in the areas of natural and artificial vision. With respect to natural vision, we cover the anatomic organization of the visual system is presented, its physiology (special attention given to receptive fields), as well as aspects of neuroscience and psychology of vision. Regarding artificial vision, we present correlated aspects such as visual information processing in linear and non-linear systems, curvature and thinning methods, as well as pattern recognition using supervised and non-supervised approaches.

Competence

Course prerequisites

Grading Philosophy

Two written and a substitutive written examinations. Several practical projects and seminars.

Course schedule

Part I: natural vision systems. 1. primitive natural vision systems (insects, arthropods, molluscs, etc). 2. advanced natural vision systems (including respective mathematic-computational modelling) 2.1. neuronal processing, principles of formation and propagation of stimulii in neutrons, respective modeling. 2.2. basic processing, retinal processing, lateral geniculate nucleous, receptive fields, superior colliculus, motor control. 2.3. visual cortex processing (neurophysiology, types of cells, modular organization in bands an pinwheels, visual cortex modelling through Hough transform). 2.4. processing in higher level cerebral structures (memory, inference, language, attention), modelling multiple stage integration. Part II: artificial vision systems (including basic principles, algorithms and implementation in sequential and parallel hardware) 1. integration between natural and artificial vision 1.1. principles of cybernetics 1.2. D. Marr�fs proposal 1.3. geometric quantized elements 2. neuronal networks for pattern recognition 2.1. perceptrons 2.2. networks based on the Hough transform 3. signal processing techinques (basic level vision) 3.1. autocorrelation and convolution 3.2. filters 3.3. the two dimensional Fourier transform 3.4. wavelet transforms 4. mathematic-computationa techniques for intermediatee vision 4.1. mathematical morphology: Minkowski�fs algebra 4.2. the Hough transform 4.3. segmentation techniques 4.4. data structures for representation of visual information 4.5. estimation of tangent fields and multi scale curvature 4.6. multiscale skeletons 5. computational models for high level vision 5.1. object oriented systems 5.2. databases and knowledge 5.3. artificial intelligence models 5.4. automatic knowledge acquisition.

Course type

Online Course Requirement

Instructor

Luciano da Fontoura Costa, Odemir Martinez Bruno

Other information

Site for Inquiry