Counting the number of banknotes in a stack using the optical technology

[Aks]

Hi People.

I am looking at a particular application. I want to build a currency
counter. The device will be based on optical technology. What I want
to do is, "look" at the side of the stack of banknotes and from there,
read the number of notes present in the stack.

There are two possible approached that come to my mind :

1.) The CD-ROM Approach

Here, the light is made to incident on the "side" of the stack and
whatever light is reflected back, is read by a device. Now, the number
of "gaps" in the light gives us the number of notes.
The upside of using this approach will be less complexity in the
electronics.
The downside will be , this approach will involve moving mechanical
parts. (In the form of a light source that 'scans' the entire width of
the 'side' of the stack)
I've found a patent for this approach. Here's the link.

Here is a list of patents related to this one
http://www.freepatentsonline.com/result.html?query_txt=PN/4694474^20 OR 4694474

And here is the US patent Number 4694474
http://www.freepatentsonline.com/4694474.html?highlight=46944742,4694474&stemming=on


To read the full texts of these patents, u need to make and account on
www.freepatentsonline.com. Its free.


2.) The image-processing approach
An image of the 'side' of the stack is taken by a CCD or a CMOS
camera. The image is then passed on to an image processing unit which
then processes the image. There can be seen visible distinct 'lines',
horizontal lines in the image. Each line representing one note. If we
can read the number of such lines, we can build such a device.
Now, I've got a patent for this as well. And en extremely detailed
one, describing, in detail, the complete hardware, software and the
diagrams of the project. Here it is.

US Patent Number 5534690
http://www.freepatentsonline.com/5534690.html?highlight=55346902,5534690&stemming=on

But being from computer science background, I am unable to understand
this completely. Can someone help me ?
I am ready to pay, if someone can make that thing for me. I am based
in India.


Ankit

 

[john jardine]

Hi People.

I am looking at a particular application. I want to build a currency
counter. The device will be based on optical technology. What I want
to do is, "look" at the side of the stack of banknotes and from there,
read the number of notes present in the stack.

Offhand I'd think you've a cat-in-hell's-chance of counting to any better
than say 60% accuracy. Look at any slab of cut paper and its impossible to
differentiate the individual items. Fancy cameras and software will not
improve on this.
For Currency, the accountants need 100% accuracy, 99.9% is unacceptable.

 

[Bob Masta]

Hi People.

I am looking at a particular application. I want to build a currency
counter. The device will be based on optical technology. What I want
to do is, "look" at the side of the stack of banknotes and from there,
read the number of notes present in the stack.

<snip>

Why are you limiting this to optical methods? The method you
propose can't read denominations anway, so why not just
weigh the stack on a sensitive scale?

Best regards,


Bob Masta

D A Q A R T A
Data AcQuisition And Real-Time Analysis
www.daqarta.com
Scope, Spectrum, Spectrogram, Signal Generator
Science with your sound card!

 

[Aks]

Why are you limiting this to optical methods? The method you
propose can't read denominations anway, so why not just
weigh the stack on a sensitive scale?

I am not interested in denominations. All I want to know is, how many
notes are there in a stack of notes. The size of the stack is of the
order of 100 notes. I want this device to be handheld and mobile. It
has to be VERY light. And this seems possible only if I use the
optical technology.


Ankit

 

[Aks]

Offhand I'd think you've a cat-in-hell's-chance of counting to any better
than say 60% accuracy. Look at any slab of cut paper and its impossible to
differentiate the individual items. Fancy cameras and software will not
improve on this.
For Currency, the accountants need 100% accuracy, 99.9% is unacceptable.

On the contrary, I think using optical technology will greatly enhance
the accuracy of the device.
The pits on the surface of a CD-ROM disk are much more fine than the
gaps between two adjacent notes, or tha width of any note. For the
record, the pits on a CD are 0.5-1 micrometer in size. Whereas a
bundle of 100 new notes (worst case) is around 7-8 mm - giving the
average thickness of one note as 7/100 = 0.07mm or 70 micrometer.
Now, if the light can read and distinguish something of the order of
1micrometer, I say, it can DEFINITELY read something of the order of
70 micrometer.
I tried taking a picture of a side of the stack of 100 notes. And even
there, I could see the notes distinctively. If you have a look at the
patents links that I provided, you will find the drawings of this.
After magnification, it becomes very clear. A good image processing
program, must then be able to 'count' the number of notes.

Ankit

 

[john jardine]

On the contrary, I think using optical technology will greatly enhance
the accuracy of the device.
The pits on the surface of a CD-ROM disk are much more fine than the
gaps between two adjacent notes, or tha width of any note. For the
record, the pits on a CD are 0.5-1 micrometer in size. Whereas a
bundle of 100 new notes (worst case) is around 7-8 mm - giving the
average thickness of one note as 7/100 = 0.07mm or 70 micrometer.
Now, if the light can read and distinguish something of the order of
1micrometer, I say, it can DEFINITELY read something of the order of
70 micrometer.
I tried taking a picture of a side of the stack of 100 notes. And even
there, I could see the notes distinctively. If you have a look at the
patents links that I provided, you will find the drawings of this.
After magnification, it becomes very clear. A good image processing
program, must then be able to 'count' the number of notes.

Ankit

In the interests of mindless curiosity I looked at a CD with a microscpe and
320X magnification. Each pit was clear and well defined, perfectly visible
as a clean, unwavering, string of high contrast white dots with black
seperators. Could even have read the binary straight off. Then looked at the
edge of a block of 20 sheets of paper (X50, X10 mag' under various light
conditions). Yes, the sheets were individually visible.
But, I found the sheet to sheet colour and edge contrast quite poor and
required the light source to be fiddled with for best results. Also required
some shuffling about of the paper to clarify edges where the fibres had
splayed and spoiled the seperation line. Can well written software tease out
more detail than an eye/brain neural net?.
Curiosity satisfied, I'll increase the cat-in-hell's chances to 65%

 

[jasen]

I am looking at a particular application. I want to build a currency
counter. The device will be based on optical technology. What I want
to do is, "look" at the side of the stack of banknotes and from there,
read the number of notes present in the stack.

seems kind of tricky..

have you condisered mechanical means?

run a phonograph needle up the stack and count pulses.

Bye.
Jasen