Scribd
Upload
17 views45 pages

CS4442 CS9542 Part 2 Lecture 4 Motion

This lecture covers motion estimation in computer vision, focusing on optical flow and motion fields, including methods for optical flow estimation like the Lucas-Kanade approach. It discusses the importance of estimating motion for various applications such as object tracking and image alignment. Additionally, the lecture introduces the Harris corner detector as a technique for motion tracking and feature matching.

Uploaded by

tiancong2013
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
17 views45 pages

CS4442 CS9542 Part 2 Lecture 4 Motion

This lecture covers motion estimation in computer vision, focusing on optical flow and motion fields, including methods for optical flow estimation like the Lucas-Kanade approach. It discusses the importance of estimating motion for various applications such as object tracking and image alignment. Additionally, the lecture introduces the Harris corner detector as a technique for motion tracking and feature matching.

Uploaded by

tiancong2013
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 45

Artificial Intelligence II

Part 2: Lecture 4
Yalda Mohsenzadeh

Slides are adapted from Olga Vesker (UW), Steve Seitz (UW), David Jacobs (UMD),
Rober Collins (Penn State)
Computer Vision
Motion
Outline
• Motion Estimation
• Motion field
• Optical flow field
• Methods for optical flow estimation
• Discrete Search
• Lucas-Kanade approach to optical flow
• Motion Tracking
• Harris corner detection
Why estimate motion?
• Many applications
• Track objects
• Correct for camera jitters
• Align images
• Special effects
Optical flow and Motion field
• Optical flow is the apparent motion of brightness patterns
between 2 (or several) frames in an image sequence
• Usually represent optical flow by a 2-dimensional vector (u, v)
Optical flow and motion field
• Optical flow is the apparent motion of brightness
patterns between 2 (or several) frames in an image
sequence
• Why does brightness changes between frames?
• Assuming that the illumination does not change:
• Changes are due to relative motion between the scene and
the camera
• There are three possibilities
• Camera still, moving scene
• Moving camera, still scene
• Moving camera, moving scene
• Optical flow is what we can estimate from image
sequence
Motion Field (MF)
Examples of Motion field
Optical Flow vs. Motion Field
• Optical flow is the apparent motion of brightness
patterns
• We equate optical flow field with motion field
• Frequently works, but not always
Optical Flow vs. Motion Field
• Motion field and optical flow are very different
Discrete search for optical flow
Computing Optical Flow: Brightness Constancy Equation

• Can we estimate optical flow without search over all


possible locations?
• Yes! If the motion is small …
• Let P be a moving point in 3D
• At time t, P has coordinates (X(t), Y(t), Z(t))
• Let P =(X(t), y(t)) be the coordinates of its image at time t
• Let I(X(t), Y(t), t) be the brightness at P at time t.
• Brightness Constance Assumption:
• As P moves over time, I(X(t), Y(t), t) remains constant
Computing Optical Flow: Brightness Constancy Equation
Computing Optical Flow: Brightness Constancy Equation
Computing Optical Flow: Brightness Constancy Equation
Computing Optical Flow: Brightness Constancy Equation
Computing Optical Flow: Brightness Constancy Equation
Computing Optical Flow: Brightness Constancy Equation
Computing Optical Flow: Brightness Constancy Equation
Computing Optical Flow: Brightness Constancy Equation

6
Lukas-Kanade Flow
Lukas-Kanade Flow
Conditions for solvability
Edge
Low texture regions
High textured regions
Observations
Errors in Lucas-Kanade
Revisiting the small motion assumption

• What if the motion is not small enough? How can


we solve this problem?
• Reduce resolution
Coarse to fine optical flow estimation
Coarse to fine optical flow estimation
Motion Tracking
Frame 1 Frame 2 Frame 3 Frame 4 Frame n

Frame 1 Frame 2 Frame 3 Frame 4


Motion Tracking
Harris corner detector
▪ Corner is an intersection of two edges
▪ They are good features to track/match
▪ Harris corner gives a mathematically representation
for this concept.
Harris Corner Detector: Basic Idea
Classification via Eigen values
Harris corner detector
𝐼𝑥2 𝐼𝑥 𝐼𝑦
• 𝑀 𝑥, 𝑦 = σ𝑥,𝑦∈𝑤
𝐼𝑥 𝐼𝑦 𝐼𝑦2
• Measure of corner response:
• 𝑅 = det 𝑀 − 𝑘 𝑡𝑟𝑎𝑐𝑒 𝑀 2
• det 𝑀 = 𝜆1 𝜆2
• 𝑡𝑟𝑎𝑐𝑒 𝑀 = 𝜆1 + 𝜆2
• K is an empirically determined constant; k =0.04 – 0.06
• Response value greater than a threshold is a corner
Corner response map
Example
Example
Example
Harris Corner Detection Algorithm
1. Color to grayscale
2. Spatial derivative calculation
3. Structure tensor setup (M)
4. Corner response calculation
5. Non-maximum suppression
Tracking Features
Tracking over many frames

You might also like

pFad - Phonifier reborn

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.


Alternative Proxies:

Alternative Proxy

pFad Proxy

pFad v3 Proxy

pFad v4 Proxy