1 / 11

16311: Introduction to Robotics Lab 2 – Robot Vision January 22, 2014 Alimpon Sinha

16311: Introduction to Robotics Lab 2 – Robot Vision January 22, 2014 Alimpon Sinha (shamelessly using old slides). 16-311 Introduction to Robotics – Lab 2, p. 1. Thresholding. Making a binary, black-and-white image. Given an image and a threshold value,

glora
Download Presentation

16311: Introduction to Robotics Lab 2 – Robot Vision January 22, 2014 Alimpon Sinha

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. 16311: Introduction to Robotics Lab 2 – Robot Vision January 22, 2014 Alimpon Sinha (shamelessly using old slides) 16-311 Introduction to Robotics – Lab 2, p.1

  2. Thresholding Making a binary, black-and-white image. Given an image and a threshold value, • Set all pixels of value ≤ threshold to 0 • Set all pixels of value > threshold to 255 (or just 1) 0 128 255 16-311 Introduction to Robotics – Lab 2, p.2

  3. Thresholding threshold(image, 75); Every pixel with value lower than 75: set to 0 Every pixel with value higher than 75: set to 255 0 128 255 75 16-311 Introduction to Robotics – Lab 2, p.3

  4. Thresholding Good for removing noise and artifacts. Of course, overdoing it would lose you precious details. 0 255 0 128 255 16-311 Introduction to Robotics – Lab 2, p.4

  5. Segmentation and Stereo Segmentation: A program that segments a binary image into connected components Algorithm – Double Raster Scan in Notes Stereo - Way of calculating depth from two dimensional images using two cameras http://www.vision.ee.ethz.ch/teaching/sada/sadalink/539.jpg

  6. What you’re doing Building a vision pipeline to estimate the distance to two targets Targets have tennis ball ‘fiducials’ in known square pattern - Can be triangulated with ‘stereo’ Steps: - Thresholding - Segmentation - Filtering / Data Association - Triangulation 16-311 Introduction to Robotics – Lab 2, p.6

  7. Example 1 2 4.2 feet! 4 3 16-311 Introduction to Robotics – Lab 2, p.7

  8. Things to Think About • How do you pick your thresholds? • How do you identify the fiducial segments? What features might be useful for this? • What are sources of inaccuracy? How do we account for them? 16-311 Introduction to Robotics – Lab 2, p.8

  9. Grading • Programming (35 points) • Threshold stage captures ball pixels - 5 points • Segmentation stage groups ball pixels - 10 points • Filtering/data association stage identifies ball segments correctly - 10 points • Triangulation stage estimates target distance to within 20% - 10 points • Additional points may be awarded for creative/impressive pipelines • Evaluation (10 points) • Take photos from prescribed distances - 5 points • Data collection procedure explained clearly - 5 points • Analysis (20 points) • Identify and explain potential sources of error - 10 points • Support analysis with experiments - 10 points 16-311 Introduction to Robotics – Lab 2, p.9

  10. Important stuff Due date Tuesday, January 28th Handin Handin a zip file of your submission on blackboard. Make sure the zip file has your source code, images, and website. Make sure your images for all pipeline stages are labeled clearly! 16-311 Introduction to Robotics – Lab 2, p.10

  11. Need help? Email • TA dlist (16311-sp14-tas@lists.andrew) • [Alimpon Sinha <alimpon.sinha@gm>] TA Office Hours • Have some scheduled for next 7 days. • In the REL • See website for the schedule 16-311 Introduction to Robotics – Lab 2, p.11

More Related