T&L Publications

When books are bound in a side binder machine, groups of pages, called signatures, are loaded into pockets, then individually picked up and placed on a conveyor to assemble the book. An operator tends the machine and refills he pockets as needed. A loading mistake can cause a whole run of books to be rejected at great loss since the machine is automated and moves fast.

What is needed is an inexpensive means to inspect the pockets to be sure the correct signatures are loaded. The pages to be inspected are like magazine pages; any color, or font, graphics or all text is possible. Typically, a system should 'remember' a sample image and then compare it to each top page of the signature as it is presented. The system must then accept or to reject them based on the match.

Yes, there are 'machine vision' systems that will inspect for proper labels, orientation and such but they are way too costly for this application since a typical side binder machine may have 20+ pockets, and each would require a system for each pocket.

One early suggestion was to use a bar coded cover sheet on the stack but that will not work since the stacks are topped off in different quantities based on work flow, and, it would be just as easy for an operator to place the wrong cover sheet on a stack.

Originally I considered something like doing a digital conversion of the video, storing it and comparing it for match, but before I go off head first, and need to learn way more than I now know, probably passing away before I ever see it done, I'll ask opinions of those that may have been there before me. I recall some time ago there was an M&M sorter that used a camera for color discrimination so maybe there is a lower cost but, perhaps somewhat less exact, system design that would work.

Any opinions, thoughts or should I just turn off my soldering stick!

Len

Lenp wrote:What is needed is an inexpensive means to inspect the pockets to be sure the correct signatures are loaded.

Place to start might be here:

http://www.roborealm.com/

Hope that helps

Vern

Vern,
I did look at the link you sent and it appears that a PC is needed for each camera.
If that's the case, to scale it up to do this job would be way out of line in both cost and complexity. Still, I think I'll dig deeper into the info!

Thanks,
Len

Any chance a small barcode or holo can be added somewhere on the first page of each packet? Easier to scan for and match on an expected pattern in an expected part of the page rather than scan & match on a whole page...

Hi Rusty,

No.
Each group of pages is called a 'signature'. These are folded in a particular order and each can have many sheets. There can be any numbers of sheets in a signature, depending on the folds, and when the collator hopper empties the workers just refill them.

'Perfect Binding', is where the pages and covers are both evenly trimmed, like a pocket novel or magazine. Since the trimming is after the binding, it would be possible to place a bar code on a waste edge, but, that would take an industry standard since the bindery gets work from many printers, or a secondary operation at the bindery to mark the signatures.

Years ago they placed 'tic' marks on the edges but that is generally obsolete now. It is a huge investment to outfit a 30 pocket collator with a smart camera at each station along with whatever computing power is required.

Be aware that the bindery does not 'own' the binding job. That is supplied by the printer and they provide only the binding service. The less bindery handling the better since if there are any damages, beyond a small allowance, they may have to pay for costly reprints to satisfy the customer order.

Good idea though!

Len

I just happened onto your post, Lenp. While this isn't a timely suggestion by any stretch, I wonder if it might be helpful.

What comes to mind is a collection of web cams, one on each pocket, with a single PC capturing images from them over the LAN. Not sure if $50-$100 per pocket is too costly. in any event, once each image is captured, a package like OpenCV could then be used to implement the templating method you mention on the captured image data one-by-one.

I looked at OpenCV as means to qualify product in a manufacturing facility I do some work for. Neat stuff! Hope this helps.