jo1776
Joined: 19 Aug 2008
Posts: 2
|
 19 Aug 2008 2:58 Re: The image of object in Brain Analog or Digital ? |
|
|
|
I agree, it is analog. This area has been studied for 30 years, funded largely by the defense departments desire to have computers analyze spy satellite photos for ICBMs during the cold war.
The real complexity of image understanding is in using database knowledge to feed processing parameters back to the camera and early stage processors.
For example, if I blindfolded you and took you into a typical middleclass home and suddenly pulled the blindfold off you would
1. quickly conclude you were in a house.
2. your brain would immediately begin to identify objects based on what is most likely to be seen in a house. You would not expect to see bushes, cars, molten steel, elephants etc. You would expect to see chairs, tables, windows, TV sets etc.
3. Having greatly reduced the number of objects you expect to see, you begin to process the scene based on what you expect to see. If you see something that looks anything like a table, you go back and reprocess it using a generalized model of what you know tables look like. They general have 4 legs, flat surface and are usually rectangular. If a 1 foot wide shadow is falling across the table and making a 1 foot portion of the edge not visible in a digital image, your brain, or advanced image processing software, will fill in that missing 1 foot line segment based on a knowledge base of what tables look like. In an image processing system you would feed new parameters down to the camera and early processors and try to find small indicates of the table being in this part of the image where you know it should be. This can be down repeatedly basing the new parameters on the results of the previous iteration.
It can be endlessly complicated. That the brain, with signal speeds of about 200 mph, can process images so quickly and effortlessly is absolutely baffling. Much is still unknown.
-jo1776
|
|
