Image Processing

Yes. You need to investigate Machine Vision Systems. However the available systems are expensive, difficult to program, and have limitations.

How many pieces total and why are they all mixed together?

Unless you a have literal tons of these to sort out every week good old fashioned human sorting is going to be cheaper than what a machine vision and related mechanical systems are going to cost you.

I am constantly amazed by some of the questions asked here.

This self-inflicted problem should never have happened to begin with.

Throw them all away and buy more of each type. The boxes they come in will have their dimensions/names/identification on the outside.

Put the new screws into bins which are labeled.

This will be much cheaper than sorting them and lots cheaper than a vision system

that no one in your organization will know how to use anyway.

To me, this sounds more like the result of some type of salvage operation. The OP is probably dissembling some type of devices and hoping to reuse or sell the fasteners. In my experience the screws are worth more as bulk scrap then they would be if sorted and inspected either manually or by some automated system. And, anyway, who would want to buy used screws?

It is difficult for me, with a literal mind, to understand different cultural and language nuances.

And typical of many lost souls who come here seeking wisdom,

harjeetsinghengg has provided no useful information.

We may never know.

Thanks every one for bringing your learning on this form.

After long time i came back on this site, one thing I notices earlier more positive replies being contributed but now more guys (not all) love to comment instead.

cutting it short.....thanks

Well,in Las Vegas used screws are very valuable,and legal.Sorting is done manually as to size,etc.

Oops! You said BUY,not rent.

My bad.

Somebody will buy them. People buy all sorts of crap. Try to sell them by the pound or dry gallon.

Actually, this is a rather common occurrence especially so when dealing with surplus stock from manufacturing overruns or large fabrication shops clearing out old surplus fastener and related stock.

A buddy of mine runs a big scrap and salvage business and it's normal to see loads of brand new fasteners dumped into buckets, barrels, bins and even industrial roll off boxes full of this sort of stuff for going out as scrap metal.

Every time I visit I make a point to try and fill a 5-gallon bucket with whatever he may have in that could possibly be of any use to me some day.

I mean hey if I can get $1000+ worth of nuts bolts screws and who knows what for $20 - $30 why not!

Wow, maybe we should start developing a vision sorting system.

Sure! I think he has at least 1000 tons of the stuff stored away.

So who wants to work on the 5+ tons of buckets barrels and bins of #4 - 1/4" machine screws/fasteners and misc?

Or the 50 - 100 or so tons of misc specialty and metric stuff that ranges from ~#10 - 1 1/2"+ and metric equivalent ranges of sizes?

I did a vision system for rejection of pizza crusts as well as fruit and vegetable sorters. It was pretty cool. this is a lot more in depth.

Let's see... 1000 tons = 2,000,000 lbs @ $0.99 lb new, since they would be seconds, that would beset say $0.50/lb = $1,000,000.00. Don't know when supply's are so let's a once a year. And this is at one location.

Well, the way I see it, first;

1.) we have get a plan of separation of material types.

2.) separate the type of fasteners.(head, threaded, screws, Etc.

2a) separate these further in size as to there dia... ....

Well you get the picture...

Yep. Once you factor it all out and still have to find a viable market what's left out of a $1,000,000 just ain't worth the work some days!

My first real job was building bucket trucks.

Every month or two, we'd clean the shop and sweep all the dropped hardware into piles and bag it.

Then, we'd all take home whatever we wanted.

Was that program called, "Encouraging Clumsiness"?

You can make something to sort anything...In this case common sense sorts it out and you sell the screws as variety packs...

A cheap way to minimise the variety would be sieving, plus magnetic separation, both above and below the sieve screens. Sure, you would have aluminum and brass and (some) stainless mixed together, but the sieve tower and magnets would reduce the variety found at each level, hence making human final sorting a reasonable request.

Separation of nuts by thread? A waste of time, recycle the mix.
Separation of screws by thread? Needs a trained eye.
Separation of screws by length? Human plus a simple gauge.

In low labor cost countries, there could be a living in it.

I've got it!

Paint each type a different color and then just sort them by color.

You're a genius lyn! LOL

I feel, that if you really want some useful answers, you might want to tell us all a lot more, why they are so mixed up for example.....

If your people are the ones mixing them up in the first place, get them stopped......as a priority.

But at this time, its simply not obvious....

Just a thought!!

Manually sorting is still your best method,but a few things to help,like sized rods to test nuts on,and a gauges for hex nuts,GO-NO GO type until the workers get trained.

After a while,they will not need the gauges.It will become automatic and effortless.

I have no problem visually sorting screw sizes fro/ 8/82 to 1 inch without measuring.

Throwing metric in with them muddies the water, so to speak,but even that hurdle can be overcome with practice.

If not a metric-english mix,it is easier.

Put the workers on production,and pay by pieces sorted accurately.

Errors will be deducted at triple the rate,so for every error,they will lose 3 pieces.

You will be surprised by how many can be sorted by a trained and motivated person.

I was thinking some type of vibratory sieve.

You have not said whether they are mixed, or just in non-labeled containers.

If mixed, then first sort for me would be mechanical. A simple vibration fed track could separate by head size and length.

Once sorted by these features, you might only have 2 or 3 styles in each lot. This will then be much easier to sort in a reliable manner.

For instance, to sort for head size you make a device with slots that gradually increase the space between each rail. Once the gap equals the head diameter, they will fall through. So your small ones will fal through close to the start, while larger ones will go further along the grid.

If you look at the "gates" used in vibrationfeeder bowls you will get some ideas for mechanical sorting deveices.

Forget the previous answers that have told you that machine vision systems are expensive and difficult to program, technology has moved on. Get a quote from a couple of suppliers for a low speed, single color (high contrast) system for distinguishing between single items passing on a conveyor. Unless you have more than one material ie brass screws, plated screws, black mild steel screws, then you could use color to differentiate. It will need to see diameter, screw length, head type and thread pitch. You need a system that will look and learn, i.e. you put a screw in front of the camera, press the teach button and tell the memory what type of screw it is looking at. Prices are now quite low, the image processing, lighting and conveyor for a system that will scan up to 10 items per second will come in at under $3000.

The expensive bit is if you need to sort the items automatically. Loading ten items per second onto a conveyor is not a practical hand operation. You will need some sort of dispenser to automatically feed the screws onto the conveyor. If there are large variations in size then pre-screening and two or three sets of less complex equipment to deal with a limited size range is usually cheaper and more reliable then a single arrangement that can cope with the whole size range.

Being able to separate into 300 different bins would be very expensive so consider running multiple passes of the same items. In the first pass you might separate into all screws of the same diameter. In the second pass you could separate all the screws of a single diameter into screws with different thread pitch, and in a third pass screws with the same diameter and thread pitch into head type, with a final pass to separate into length. You may get away with as few as fifteen separation gates at 5% the cost of doing everything in a single pass. Most modern vision systems could cope with changing between the various scan criteria at the touch of a button, with no additional capital cost. Fast acting separation mechanisms cost a lot more than slow acting separation mechanisms. Compromise as much as practical on speed to keep costs low.

At $3000 that's not bad if it can do basic multi-bin (2 - 6 bins) sorting at any reasonable rate of a basic conveyor type feed system.

If a person had to do a ton or so of small assorted hardware every week I can definitely see the advantage that price.

Now with any irregular volumes less than that hiring basic human labor would likely still be the way to go.

Oddly enough there are people out there that love that sort of mind-numbing repetitive tedium of sorting that things like. One of my good friends is that way. Give him $20 and hour and he will sit at a table and sort hardware by hand until he goes blind or his fingers fall off.

Personally, if its cold windy miserable day I rather enjoy sitting at a table doing similar things now and then as well.

Can we solve this problem through image processing

Yes but the solution will be orders of magnitude more expensive than the product if this is just a one-off or infrequent application.

What's the application exactly? Are you getting screws supplied mixed for some reason? Just how many batches of screws are you looking at? What types of screws (are they different materials, sizes, etc)?

More information would help to narrow down viable solutions. Image processing is only really viable for large quantity mass production, and generally as a quality check to ensure products don't accidently get mixed, not to sort a pile of mixed together product.

why not weighing based?

I recently joined CR4 and saw this post. To break down the vision world there are smart cameras and GIGE cameras. There are 2 main industry leaders when it comes to "smart" vision systems: Keyence & Cognex. They are both relatively comparable in pricing. We have used both systems at our facility doing screw measurements and sorting. We have strayed away from Cognex in an effort to keep up with our line speed. They Keyence cameras are much more capable when it comes to processing power. I would say depends on what your goals are moving forward. I would take a look at both companies anything vision related. Cognex usually operates through distributors, commonly with some sort of integration/controls group which can be a benefit when looking for turnkey solutions. Keyence is the Cadillac of vision though and provides a range of different automation products which may be able to reach your needs. Just my 2 sense working with a handful of different vision systems and automation techniques.

so, what separates Keyence and Cognex is processing speed, do you have any comparative literature on each?

Is there any other differences.