Problem with a Loom!

[ericgibbs]

hi xanadunow,

Are you directly involved with or representing the sack manufacturer concerning this project or are you just trying to answer
the opening posters thread made by 'Wizard'.?

 

[xanadunow]

no connection

I am in Oz ericgibbs. Wizard is in the US.. I am just passionate to determin resonably cheap and workable solution. Is it not what this forum is all about?

Regards,
xanadunow

 

[ericgibbs]

I am in Oz ericgibbs. Wizard is in the US.. I am just passionate to determin resonably cheap and workable solution. Is it not what this forum is all about?

Regards,
xanadunow

hi xanadunow,
Its not a criticism, its just the way your posts are written mislead me into thinking you were actually making decisions on the way the project would be done.

I agree, the forum is an excellent venue for developing ideas and suggestions.

 

[Ubergeek63]

There is 1 sec of time.

There is "x" times "y" pixels in the frame..

Can you make the decision using these PiCs in 1 sec?

Regards,
xanadunow

I have to say no. The fabric grows by some multiple of 1-2" per second. There are 640x 480 pixels in a frame. There are 3 bytes to process per pixel. The contrast is lousy. The background level is inconsistent.

In short you are proposing a tremendously complicated solution to something that would be easy if they would submit to changing batteries a few times a year.

Dan

 

[xanadunow]

Eric

Hi Eric,

You are the 2nd person to "pick me" up on that. I guess, I am coming to strong in my beliefs (or to stupid).

As naive as it may sound, I do trust and believe in the end result. I did arrive on this forum with a single (not answered as yet - question). I felt that I do owe something to the forum and have decided to share my experience, knowing - that being in the need of your help - does put me in debt I can only repay by offering my own experience I can share - in kind.

Please, do correct me - if I am wrong.

I will say it again, I have no connections - just the desire to find the solution.

Regards,
xanadunow

 

[xanadunow]

Dan

OK Dan, I do accept it. You did trash that idea for me when I felt it may be the best. OK, back to the drawing board then, and - your batterries

Regards,
xanadunow

 

[Ubergeek63]

OK Dan, I do accept it. You did trash that idea for me when I felt it may be the best. OK, back to the drawing board then, and - your batterries

Regards,
xanadunow

Sorry... I have designed a few video displays and we are always looking to back off on the bandwidth. All the customer "knows" is "more is better".

Dan

 

[xanadunow]

I do feel very comfortably with your assessment Dan, I did get lost following my arrival at the "xy" frame, having the hope only that it could be "processed" within the 1 sec of time space, but not knowing the exact method of doing it, and - seeing the pitfalls of my luminance sampling I did hope that some of you guys can "jump in" and improve on this approach with a conceivable outcome.

Regards,
xanadunow

 

[Ubergeek63]

I do feel very comfortably with your assessment Dan, I did get lost following my arrival at the "xy" frame, having the hope only that it could be "processed" within the 1 sec of time space, but not knowing the exact method of doing it, and - seeing the pitfalls of my luminance sampling I did hope that some of you guys can "jump in" and improve on this approach with a conceivable outcome.

Regards,
xanadunow

I am not even sure off hand how long it would take on a PC.

An actual machine vision system has both the power and customization for it. Stuff like high contrast video with just enough resolution to allow reliable detection. We are assuming a low cost solution with no overkill needed.

Dan

 

[xanadunow]

the pattern

I was thinking more of recognising the difference between the correct pattern:

101010101010
010101010101
101010101010
010101010101

against the pattern with one string missing:

101010101010
010101010101
010101010101
101010101010

Each "0" represents here a "visible" horizontal string overlapped by a visible "1" vertical string.

With a missing string, the "vertical" overlap will capture not one but 2 horizontal strings and this is all I would like to detect to make the decision.

In fact, any deviation from the original pattern as at the very top, should stop the line. This would cover the case when more than one of the 4 horizontal strings is missing or has ended.

I am just not sure how to implement it Dan..

Regards,
xanadunow