ECE 549 - Computer Vision

Semesters Offered

Official Description

Information processing approaches to computer vision, algorithms, and architectures for artificial intelligence and robotics systems capable of vision: inference of three-dimensional properties of a scene from its images, such as distance, orientation, motion, size and shape, acquisition, and representation of spatial information for navigation and manipulation in robotics. Course Information: Same as CS 543. Prerequisite: ECE 448 or CS 225.

Prerequisites

Credit in CS 225 or ECE 448

Subject Area

Robotics, Vision, and Artificial Intelligence

Course Directors

Description

Examines information processing approaches to computer vision, and algorithms and architectures for artificial intelligence and robotics systems capable of vision: inference of three-dimensional properties of a scene from its images, such as distance, orientation, motion, size and shape, acquisition and representation of spatial information for navigation and manipulation in robotics.

Notes

Same as CS 543.

Topics

  • Image formation: imaging techniques - monocular, binocular; range, brightness, color; digital images
  • Early vision: reflection and brightness; edge detection; human stereo vision, range from stereo images; surface orientation from shading; motion perception in humans; motion estimation from optical flow and motion stereo; motion from optical flow; human texture perception and surface shape from texture gradient; surface shape from contours; active vision
  • Segmentation: multiresolution and multiscale image representations; segmentation; texture models, recognition and discrimination, texture based segmentation; segmentation; texture models, recognition and discrimination, texture based segmentation; segmentation from motion
  • Representation and robotics: image boundary and region representation, shape description; three-dimensional surface representation, object and viewer-centered representations; dynamic representation; navigation
  • Matching and recognition: three-dimensional model generation; object recognition; relaxation processes
  • Computer architectures for vision: model of computer vision; planar array and hierarchial multiprocessor architectures for image analysis and processing

Detailed Description and Outline

Topics:

  • Image formation: imaging techniques - monocular, binocular; range, brightness, color; digital images
  • Early vision: reflection and brightness; edge detection; human stereo vision, range from stereo images; surface orientation from shading; motion perception in humans; motion estimation from optical flow and motion stereo; motion from optical flow; human texture perception and surface shape from texture gradient; surface shape from contours; active vision
  • Segmentation: multiresolution and multiscale image representations; segmentation; texture models, recognition and discrimination, texture based segmentation; segmentation; texture models, recognition and discrimination, texture based segmentation; segmentation from motion
  • Representation and robotics: image boundary and region representation, shape description; three-dimensional surface representation, object and viewer-centered representations; dynamic representation; navigation
  • Matching and recognition: three-dimensional model generation; object recognition; relaxation processes
  • Computer architectures for vision: model of computer vision; planar array and hierarchial multiprocessor architectures for image analysis and processing

Same as CS 543.

Texts

Forsyth and Ponce, Computer Vision, Prentice Hall, 2003.

Collateral Reading:
B. Horn, Robot Vision, McGraw-Hill.

Last updated

2/13/2013