unisa ITA  unisa ENG


courses

Vision-based Measurements ("Misure basate sulla Visione")

Objectives: The course treats topics of major interest related to measurement systems and methods based on the analysis of digital images adopted for the automation of industrial processes. On-line and contact-less systems and methods for vision-based measurements (based on image processing) of dimensional parameters of parts and products will be described, and models related to the evaluation of measurement  uncertainty of the results will be introduced.

  • Credits: 6
  • Type of course: optional
  • Year: 2nd year, 2nd semester of the Master of Science in Electronical Engineering, University of Salerno, Italy
  • Lecturer: Dr.Paolo Sommella
  • Language: the official language is Italian, although English can be used for questions, consulting hours, additional notes and material
  • Examination: oral, consisting in a discussion of a final project + a discussion of a topic of the course

 

The presentation of the course for the academic year 2012/2013 is available. 

 

Syllabus

  • Introduction. Introduction to measurement systems based on image analysis.
  • Devices. Cameras for industrial applications: Architecture and features. Introduction to optics. Thinlens model and relationships for the design of optics. The depth of field and the diffraction. Illuminators and their main characteristics. Types of light sources for industrial inspection and criteria for an optimal choice.
  • Analysis of digital images. The digital images. Software architectures for image-based measurements. Point transformations: histogram and its applications, thresholding, contrast enhancement. Local transformations: linear and nonlinear types. Edge detection. Canny 's algorithm. Hough transform. Numerical exercises in LabView, Matlab and C/C++.
  • Camera model and calibration. The process of image formation. The "pin hole" model. The equations for aperspective projection. The distortion of the lens. Direct Linear Transformation algorithm (DLT) for camera calibration. The calibration target. Decomposition of the perspective projection matrix. Correction of lens distortion. Propagation of uncertainty through the DLT calibration algorithm. Numerical exercises LabView, Matlab or C/C++.
  • Stereo measurement systems. Measurement systems based on stereo vision. Measurement of the depth in the case of simplifying assumptions, and evaluation of its uncertainty. Stereo reconstruction in the general case with linear methods, and uncertainty propagation. The epipolar geometry. Numerical exercises in LabView, Matlab or C/C++.
  • Final project. In the final part of the course (approximately during last four weeks) students develop a measurement application based on the analysis of digital images and on camera calibration. The development of software component of the project will be done in a programming language to be agreed (typically in LabVIEW or C/C++ using OpenCV libraries, but other choicesare possible as well).

 

Couse Books

Other books and lecture notes can also be used. Please contact the lecturer for any question.

  • R.C. Gonzalez, R. C. Woods, "Digital Image Processing", 2nd Ed., 2002, Prentice Hall, Inc.
  • Alfredo Paolillo, "Appunti di Misure basate sulla Visione", Lecture notes (in Italian), downloadable here.
  • R. Hartley, A. Zisserman "Multiple View Geometry in Computer Vision", 2nd Ed., Cambridge University Press.

 

downloads