Interesting web site. Here is strategy to consider.
- copy the flume/sluce because they got it to work.
Reproducing it will show you why.
- put a detector array on the exit wall and connect it
to a parrallel port connector
- get a device driver to read parallel port and write it
to a file.
- experiment with some fish and find out what kind of
data pattern your recording.
- then write routines that sample and process the data
for recognition of a passing fish. It will be hard to code
a solution before you know the data pattern.
- get it to accurately count one fish then add more
Good Luck
"It seems the harder I work the luckier I get."
( for got who said it)
Larry :-)
On Friday 20 December 2002 02:28 pm, Chuck wrote:
> Larry,
> My comments below...
>
> > -----Original Message-----
> > From: Larry Sanders [mailto:rhatman@earthlink.net]
> > Sent: Friday, December 20, 2002 08:34 AM
> > To: slug@nks.net
> > Subject: Re: [SLUG] A Challenge....
> >
> >
> > Chuck
> > Remember the magic words: "all you gotta do" !
>
> Yes like the "oh yea that is easy"
>
> > Here are a few engineering questions.
> > What are the dimensions (HxLxW) of the pinfish and pigfish?
>
> These guys can range from 1X2.5X.5 to 3X7X1.5 inches
> though we delivered one last night that was really big enough
> to feed a person and leave them full.
>
> > This is the area that will interrupt the IR beam.
> > Does the IR detector signal on/off or an analog signal?
>
> I can go either way with this one, I figure a ON/OFF is
> probably the best way as it will not require a A/D conversion.
> I have also thought about using bluegreen to blue LED's with
> sensors optimized for that part of the spectrum because it
> is water frendly.
>
> > If it is an analog signal then how will you do a D/A converter?
> > This is a key question if the spacing between the detectors is
> > significantly less than the size of the fish.
>
> Yes, from what I have seen of the Icelandic product as the fish
> moves through the sensor it "images" the fish and then looks at
> the "image" to figure out if it is a fish. Also it checks to see
> if there is another fish "tailing" the first one just in case
> the second one has pushed up beyond the tail of the first one.
>
> > Is the sluice also pouring water or just allowing the fish to swim?
>
> yes, the sluice is a "V" shape, it is inclined downward at a angle
> such that the fish follow the flow of water into the sluice which
> narrows down to the exit point which is where the sensor is located
> This address is the web page where this counter is located
> http://www.vaki.is/Forsida/Enska/Menu.htm
> There are 4 images on the screen, the upper lefthand one of the fish
> exiting the sluice is the one to click on, you can see more detail
> on that page.
>
> They also have a image system that counts fish in a tank but it is
> for even more volume.
>
> > If you are counting from a verticle view, then the sluice should be
> > shallow and the detectors close togeather. Or if the detectors are
> > horizontal then the sluice can be narrow and possibly deeper.
> > It will be less complicated if the water flow is zero and the fish are
> > swimming between the two tanks. Then you can get them to
> > line up like a cattle gate.
>
> The detectors would be vertical or on leg of the 'V' at the mouth
> of the sluice.
>
> > In one post you mentioned that you would like to be accurate
> > within 5 fish. Five fish out of how many? What percentage?
>
> I should have given percent, %98 would be a good target to shoot
> for, I could probably get away with %95. We only get 25 cents/fish
> so we can not give away many, and at the same time we do not want
> to shore change the shops or they will go elsewhere.
>
> > The problem of counting has been asked before in the field
> > of industrial engineering. Significant time invested in a search
> > for publishd papers and solutions will definately be worth it.
>
> Most of the solutions are for fresh water fish, and use a
> tube with a set of 3 metal rings inside the tube, as the fish
> swim past the rings the water conductance changes registering
> a count, these devices are not expensive running about $200-400.
> The optical counter system which will handle salt water fish
> is nother issue, at $5250/unit it is a bit steep for counting
> pin fish.
>
> > The key to this is the D/A converter and the corresponding
> > signal processing. The sample rate should be 100 times the
> > rate of the fish in order to recognize a pattern the typical
> > pattern of expanding/decreasing shaddow to seperate two
> > fish head to tail. If a fish swims by at 1 per second then a
> > sample rate of 100/sec or 1 in 0.01 seconds is not to fast
> > to process with today's cpu's. The software will ask at each
> > sample: is this shadow getting bigger or smaller?
> > All you gotta do is count the fish.
>
> That is a very good sum of it, and now all I have to do is come
> up with the software part, hardware I do not see trouble with.
> I have both PIN diodes and photo transistors, there are also
> linear arrays of these things that would do exactly what I
> want to do on the hardware side. Now just to figure out how to
> to it.
>
>
> *****************************************************************
> This e-mail and any files transmitted with it are confidential and are
> intended solely for the use of the individual or entity to whom it is
> addressed. If you have received this transmission in error, please notify
> the sender immediately and destroy any hard copies you may have printed and
> remove all copies of the e-mail from your hard drive. Opinions, conclusions
> and other information in this message that do not relate to the official
> business of Utility Partners, Inc shall be understood as neither given nor
> endorsed by it.
>
> Visit us on the web at http://www.utilpart.com
> *****************************************************************
This archive was generated by hypermail 2.1.3 : Fri Aug 01 2014 - 19:45:38 EDT