harshad's questions:

HARSHAD: but is 32x32 resolution enough?? i mean will it be able to give them enough information?

HARSHAD: i converted a 640x480 to a 32x32(approx. one dim should be greater than 32 according to the aspect ratio) and i saw that we wont be able to identify distinct objects...they kind of merge with each other..

so what i want to say is...the whole point of this project is to make the blind capable of identifying distinct objects in front of them rite?
now if i cant identify them in the picture with my eyes...how do u expect them to identify them by just listening to it..

i mean for them to identify the objects...the program must be able to do it.
in my opinion...only then, there is some point. it will be very difficult(almost impossible) if the program just plays the pixels column wise.

please enlighten me, how do u tackle this problem?

Harshad:
so, i dont say the program tells them what object is in front and what is its colour...(i read a thread on homepage, that they want to be independent, and i agree with it)

im saying, what struck me, that the program acts as a radar. so when this happens..the pixels and all dont matter..im not a good programmer(learned a lot of C++, but not any other language..) so i dont know how much this is feasible..

but what the program does is scans the image from say left to right..and gives a beep when it thinks it encountered an object.



OMEE:


pictures speak better than words


32x23 pixel images are good enough for navigating ones own way though this world.


sunset - 18 x 25 pixels

Kylie Minogue 35 x 35 pixels picture

Dolphin 32 x 32 pixels

All these images are below 64x64 pictures...

a baby's smile... 36 x 46 pixels

Baby #2 49 x 45 pixels

wanna go up stairs?
35 x 50 pixels (Stair Case #1)

38 x 50 pixels (Stair Case #2)

Want to see far away scene as tiny image??
49 x 50 pixels (HUGE Hollywood Letters on the hill)


40 x 60 pixels (Laetitia casta)

40 x 50 Pixels (josie maran)


Hollywood Stars:
Tom Cruise 40 x 50 pixels picture


60 x 40 pixels (Mercedes Car)


conversion messes up the image.. good algorithms are needed to scale down the image. else.. like u said.. objects merge.

the idea is NOT to give them vision as best as we do right now.
thats highly unfair.. cuz we spent 5 -6 years of our childhood to develop our vision and intensive learning.

its very illogical to expect these peoples brains to learn/ be able to identify images in few short days just like the sighted people have learnt over years to do!

also for the sighted people who have already developed and matured the image sensing.. its very difficult (but not impossible) to learn this language. where as for those who have no other option.. its quite easier. especially for those who have never seen light/ color.




radar thing as well as black and white image sonification is already successful.. google it to get more info.

auxel is all about coloured images! and that too with clarity!!

more questions are welcome...

cheers

omeee