CSCE 181 Guest Lecture 7
Jump to navigation
Jump to search
« previous | Thursday, March 31, 2011 | next »
Overview of Computer Vision
Science and technology of machines that see
Obtain information from images, video sequences, multiple cameras, scanners, etc.
Applications
- Robot perception for better autonomy
- Augmented reality (labeling real-life objects with HUD)
- Detecting events (visual surveillance, counting objects)
- Modeling objects/environment (create a 3D counterpart to the real world)
- Facial recognition
- Computer interaction (automatic login, games)
- Organizing Information (indexing images and videos in databases like on Flickr and YouTube)
Image Representation
2D array of pixels: void[][]
- Grayscale
- intensity (0–255; 8 bits/px)
- RGB Color
- 3 Channels (0–255; 24 bits/px)
Image Interpretation
From Single Images
- Edge Dectection
- Corner and feature detection
- Geometric Shapes
From Multiple Images
Parallax from 2 images used to estimate depth of scene
- Reverse: use 2 images to estimate position of camera
- Find pixel in one image
- Find same pixel in another image
- measure distance between them
Footnotes
- Open source Computer Vision Library: OpenCV