Objectives

The projects for the subject "Vision and Industrial Perception Systems" are the following:

  • Dimensioning, utilization and application to new situations of perception systems at distance (specially artificial vision systems);
  • Understanding of the image formation and the principles of treating (processing) the digital image;
  • Extraction and evaluation of geometrical and visual properties without physical contact of measuring tools;
  • Acquire capabilities of elaborating programs (prototypes and industrial) to solve perception problems.

    • Project - Recognition of game pieces

    This project required a MATLAB script to process a set of generated random pictures with added noise containing game pieces. The parameters to detect were:

  • Number of domino pieces;
  • Number of dice faces;
  • Number of game cards;
  • Number of dominos facing determined direction;
  • Number of dice faces facing a determined direction;
  • Number of domino pieces with the same number of dots on both sides (doubles);
  • Total accumulated number of dots from dominos and dices;
  • Number of cards from determined suit;
  • A string with the digits of the number of dots of all cards in ascending order.

    • Project - Recognition of biscuits

    This project required a MATLAB script to process a set of generated random pictures with biscuits as elements. This project contains different colour backgrounds in every image and many of the elements to identify are broken. Parameters to detect were:

  • Number of elements interesecting the image border;
  • Number of total elements that are broken;
  • Number of total elements that do not intersect the border and are not broken;
  • A list of non broken elements by number of biscuit type.

    • Project Gallery

    This gallery contains some examples of the pictures that needed to be processed for each project. The program would run sets of pictures and return a text file with the parameters of each picture.

    Example of Industrial Perception laser scan

    The machine uses a laser to retrieve information to be processed.

    Project - Game Pieces Recognition

    Generated Image: Example 1

    Project - Game Pieces Recognition

    Generated Image: Example 2

    Project - Game Pieces Recognition

    Generated Image: Example 3

    Project - Biscuits Recognition

    Generated Image: Example 1

    Retrieved Relevant Object Properties/Descriptors

    The four most important object properties for this project: Form Factor, Extent, Solidity and Eccentricity.

    Project - Biscuits Recognition

    Generated Image: Example 2

    Project - Biscuits Recognition

    Generated Image: Example 3

    Conclusion

    This subject provided important knowledge in the automation area specialized in vision systems. It was understood how machines percept images and how it is possible to expand the current limitations of image processing using MATLAB's digital image processing toolbox. Finally, writing scripts for image processing of sets of pictures helped honing the necessary skills for developing foul proof algorithms.

    MATLAB Programming

    Scripts and Algorithms - Digital Image Processing

    Autonomy

    Taking iniative to study, develop and test different ideas on my own.

    Data Processing

    Process Matrices of data in MATLAB and results analysis.