Open Problems in Computer Vision(oProCV)
Topic 4. Motion
MSc. Luis A. Mateos
Open Problems in Computer Vision(oProCV)
Motion
detecting motion
optical flow
tracking single objects
shadows and occlusions
dynamic scene representations
Open Problems in Computer Vision(oProCV)
Motion
Motion means a change in the location of a body.
A body which does not move is said to have
constant (time-invariant) position.
Motion is always observed and measured relative
to a frame of reference.
Everything in the universe is moving[1].
Motion signifies any spatial and/or temporal change in a physical system.
[1] De Grasse Tyson, N., Liu, C., & Irion, R. 2000. One Universe: At home in the cosmos. p.20–21.
Joseph Henry Press. ISBN 0-309-06488-0
Open Problems in Computer Vision(oProCV)
Motion Detection
Motion detection quantifies the motion of a moving object within the field of view.
Algorithm by a fixed camera compares the current image with a reference image and
simply counts the number of different pixels.
Images will differ due to factors such as varying lighting, camera flicker, etc..(noise),
so pre-processing is needed to reduce the number of false positive.
Open Problems in Computer Vision(oProCV)
Optical Flow
Is the pattern of apparent motion of objects, surfaces, and edges in a
visual scene caused by the relative motion between an observer and the scene.
Try to calculate the motion between two image frames which are taken at
times t and t + δt at every pixel position.
For a 2D+t dimensional case (3D or n-D cases are similar) a pixel at location (x,y,t)
with intensity I(x,y,t) will have moved by δx, δy and δt between the two image frames,
Open Problems in Computer Vision(oProCV)
Optical Flow
Every arrow from the optical flow is a vector, defined by a magnitude and an angle.
Open Problems in Computer Vision(oProCV)
Optical Flow
Open Problems in Computer Vision(oProCV)
Optical Flow
Open Problems in Computer Vision(oProCV)
Optical Flow
Open Problems in Computer Vision(oProCV)
Tracking single objects
Considerations
Type of object:
- Rigid
- Deformable
How to isolate the object from:
- The background
- Other objects
- Noise (Ilumination, camera,)
Open Problems in Computer Vision(oProCV)
Tracking single objects
Common algorithms
Blob tracking:
Segmentation of object interior (for example blob detection, block-based correlation or
optical flow).
Kernel-based tracking (Mean-shift tracking):
An iterative localization procedure based on the maximization of a similarity measure.
Contour tracking:
Detection of object boundary (e.g. active contours or Condensation algorithm)
Visual feature matching:
Registration
Open Problems in Computer Vision(oProCV)
Shadows and occlusions
Shadows
A shadow is an area where direct light from a
light source cannot reach due to obstruction
by an object.
It occupies all of the space behind an opaque
object with light in front of it.
The cross section of a shadow is a
two-dimensional silhouette, or reverse
projection of the object blocking the light.
Open Problems in Computer Vision(oProCV)
Shadows and occlusions
Occlusions
Partial occlusion, when the object hide only part of it from the camera.
Open Problems in Computer Vision(oProCV)
Shadows and occlusions
Occlusions
Complete occlussion, when the object is hidden, in the current frame it cannot be found.
Open Problems in Computer Vision(oProCV)
Dynamic scene representations
One of the more powerful impressions created by vision is that of a coherent, richly
detailed world where everything is present simultaneously.
Indeed, this impression is so compelling that we tend to ascribe these properties not
only to the external world, but to our internal representations as well.
For example, changes in images of real-world scenes often go unnoticed when made
during a saccade, flicker, blink, or movie cut.
This "change blindness" provides strong
evidence against the idea that our brains contain a picture-like representation of the
scene that is everywhere detailed and coherent.
Open Problems in Computer Vision(oProCV)
Dynamic scene representations
How then do we represent a scene?
It is argued here that focused attention provides spatiotemporal coherence for the
stable representation of one object at a time.
It is then argued that the allocation of attention can be co-ordinated to create a
"virtual representation".
In such a scheme, a stable object representation is formed whenever needed,
making it appear to higher levels as if all objects in the scene are represented in
detail simultaneously[2].
[2] The Dynamic Representation of Scenes
Author: Rensink, Ronald A.
Source: Visual Cognition, Volume 7, Numbers 1-3, 1 January 2000 , pp. 17-42(26)
Publisher: Psychology Press, part of the Taylor & Francis Group
Thank you