Search Here

MISC
 Parts List
 Robot Forum
 Chat
 Member Pages
 Axon MCU
 Robot Books
 Shop
 Contact

SKILLS
 How To Build
  A Robot
  Tutorial
 Calculators
 Mechanics
 Programming
 Miscellaneous
 Robots
 Space

HARDWARE
 Actuators
 Batteries
 Electronics
 Materials
 Microcontrollers
 Sensors

SCIENCE
 Robot Journals
 Robot Theory
 Conferences

PROGRAMMING - COMPUTER VISION TUTORIAL


Hi,

I was just browsing through your pages on computer vision and hat a couple of comments.

Firstly you have this comment on frame rates:

"This is also quite similar to how a human processes temporal information, as we see about 25 images per second - each processed individually (I couldnt find a reliable reference to verify this claim, contact me if this is wrong!)."

I think it is wrong but mostly in the way biological vision is often compared to and explained in terms of electronic cameras and video systems. It is often misleading to place biological vision within the idea of 'frames' and 'frame rates' used in electronic systems.

It is true that our vision appears to work at about 25 FPS but this is only apparent because we do not tend to perceive changes in the visual field that are above this rate. This is because of the response rate of the vision system to changing stimuli (in crude terms very rapid changes get averaged out) In terms of the system behaviour biological vision deals with constant multiple and parallel channels of incoming signals that work their way (in parallel mostly) through various neural processing systems. Some of these systems can respond to very rapid fluctuations in stimuli and others cannot but there is no 'frame rate' for the brain.

The most obvious evidence for this is what happens when you are exposed to oscillating stimuli at a rate close to 25hz. If we worked with frames then we would frequently perceive interference patterns as our 'frame rate' interfered with the stimuli 'frame rate', especially when the two were almost in sync. It is true that some flashing stimuli can cause fits in certain people but as far as I know this is down to an unfortunate synchronisation effect of neuronal firing patterns in response to the flashing.

(edit of unrelated info here)

Nice site by the way ;)

Bill

<<< return to the Computer Vision tutorial

Get Your Ad Here