FAQs: Digital Perforation Reader (DPR)

The first thing to understand is that NEW DPR Technology offers a tool not a solution, however – applied pro-actively the technology uses advanced Facial Recognition Technology (FRT) to accurately measure and identify stamp perforation ​

The following FAQs may assist. We will be actively revisiting FAQs to revise/update as DPR evolves utilising machine-learning and AI: –

IMPORTANT: PLEASE NOTE – Your DPR Card works best placed upon a light background such as a blank sheet of white paper

Because your DPR is effectively a ‘hand-held’ instrument … just like a camera – with varying light/focus – results vary, however:

DPR is an interpretive instrument, a philatelic tool which used intelligently can produce the most remarkably result accuracy.

Initially in trials, we experienced significant result fluctuation. DPR members may find that over the course of a few ‘perfing sessions’ measurement seemingly improved almost without additional input – basically because the DPR user has started – just like getting used to working with a camera … to automatically eradicate variable factors by working with the known, rather than the unknown.

How to get the best from DPR technology?:

THE JOURNEY

Initial use: Review animation – https://www.upastampauctions.co.uk/dpr-picture-line/

Using QR code on DPR card/internet – Access DPR web-app on your mobile/cell phone/iPad/PC

Review FAQ’s for 1-2-3 click DPR algorithm analysis:

https://www.upastampauctions.co.uk/upa-stamp-perforation-gauge-user-guide/

 

DPR offers an interpretative ‘auto-tool’ technology enabling collectors and dealers to interpret/attribute results.

• Whilst it is not essential – Please search Google for your mobile /cell phone make/model + HOW to turn off camera ‘auto-rotate’

 

The FIRST thing to learn about perforation is that Stamp Catalogue listings were never accurate right from the word go, when 100+ years ago stamps were itemised/allocated/listed by perforation – 

Logic seems to confirm this because British Stamp Printers produced many of the early classic stamps for other countries / colonials … and presumably post 1840 these printers were working in inches –

so that stamps produced until 1970 may have been measured in increments of ‘inches’ … or perhaps printers turned to MM / CMS measurements prior to British LSD-Decimalisation in 197l … an interesting subject

to research especially as inches/feet/yards/miles are still very much ‘common-currency’ in the UK today, as are acres … 

When we started testing there were many tweaks and edits upon our journey; surely as we build and refine DPR there will be many more …

Sometimes we would achieve wide-ranging variables.  Other times we might achieve ‘error messages’.

We soon discovered that although  (FRT) Facial Recognition Technology may work within a margin of error with ‘grainy’ images … similarly if your phone camera cannot record a legible image then DPR cannot read perforation.

In practice … everybody knows that no two hand-held photographs will be alike due to lighting/angle/framing factors/conditions … effectively DPR is no different because with light/focus/framing variables the image(stamp) is captured by hand-held digital camera (mobile phone).

• Overriding DPR Issues are therefore: Focus, Flat stamp, Consistent one direction light eliminating shine/conflicting light (Turn Flash OFF), Consistent background contrast.
• It felt that almost like when I was learning to wind-surf for the first (and only) time – after falling in dozens of times … that after repeated frustration a ‘sweet-spot’ of practised use emerged: so it was with DPR. Although DPR is obviously technology, so it is also about technique.

As we worked toward DPR production we realised that we had naturally maximised results by creating consistency in daily use, so that looking through the camera viewer – the stamp, laid flat upon a consistent matt black (single source lighting) with minimal conflicting lateral(sideways) light  produced best results:

Whilst working with DPR we had found the best optimised position in our stamp office (often in the darkness of evening) produced accurate results. In a sense this bears similarity with use of the office philatelic ‘Ultra-Violet’ lamp which does not like bright light. Also upon testing the £300+/- ‘Perfotronic’ machine which could not cope near the office window. Each of these devices, even the electronic watermark detector … EACH requires optimised location in the office to facilitate best result.

DPR is sensitive …

1. Obscured perforation: We soon discovered that if the perforations upon any side of the stamp being ‘perf’d’ were obscured then an error message would be generated,

Fundamentally DPR needs to be able to ‘read’ perforation upon ALL four sides.

1. Matt black background: Early DPR test cards were produced upon shiny plastic card. We discovered that shiny black plastic acts as a ‘receptor’/bouncer for light: random light shining/bouncing across the surface of DPR cards invalidated results.

Consequently, by testing different materials, eventually we alighted upon paper-fibre biodegradable finish matt black background cards like the card you hopefully may keep in your wallet when you feel comfortable.

By Eradicating/eliminating background ‘noise’/light reflection we were able to refine to accuracy levels corresponding to catalogue-listed perforation.

1. The use of ‘flash’ has never produced a readable result.
2. Focus needs to be accurate upon both DPR card and stamp. DPR uses the ratio of DPR card to stamp image upon it to diagnose/read result. DPR cannot read blurred images.
3. Later testing determined issues with perspective … moderate perspective correction was configured within the algorithm so that perspective has been accommodated.
4. DPR Card works best upon a neutral/white contrasting background.

Finally:

1. Naturally – Curling/bending stamps produce incorrect results. We place a clean small minimally reflective coin/minor object (e.g. cut-down pencil eraser rubber is excellent), centrally well clear of perfs to flatten a curling stamp … works a treat.

Talking about technique, upon reflection, I realise that I probably focus DPR card and stamp in the same area of my camera viewer screen

Re: Images below taken 23/12/2024:

Notes:

1. DPR readings: see images/compare results immediately below:
2. Yellow: Manual readings extracted/confirmed by SG ‘Instanta’ perforation gauge
3. Note image 2: Mobile camera has auto rotated image readings accurate but transposed

Some other FAQ’s

Cautionary Note: Everybody knows that mobile cell phone /smartphone advanced technology devices are designed to assist users by rotating their screens automatically from vertical to horizontal based upon movement. For this reason, it is wise to check that your DPR results match catalogue listings. Never take it for granted that you have found a different perforation which is not catalogued without corroboration – your results may simply be transposed by your mobile phone cameras sensitive sensors.

DPR Do’s & Don’ts

Resolving a DPR Member’s 1^st issues.

Please refer images further below:

We’ve placed images side by side: this allowed us to advise what may be impacting a DPR Members first results.

1. Your DPR card should best be horizontal and the camera vertical as indicated in the online animation – Please refer images below – Nb: DPR Administrator image on Right Hand Side
2. Your stamp should be upright
3. As you see below the DPR Administrator card (RHS) is ‘skew-whiff’ … but not so much as your DPR Member image (LHS)
4. ‘Perspective-correction’ permitting skewed images has been built in to DPR facial recognition technology … but the image LHS is extreme. Extreme images impact results negatively.
5. Basically the DPR Member User image below (LHS) generating the incorrect reading (centre image) has been taken with the mobile phone tilted or dipped.
6. The DPR Administrator Image below (RHS) indicates that the mobile phone camera was quite level – consequently creating reduced perspective thereby generating an accurate result.

• The most important thing other than ‘framing’ and level device/level stamp is not to have extraneous light crossing the surface of the DPR card when taking the photo.
• Just as in photography through user testing and experience we have acquired the knack to look for random light crossing the surface of the DPR card and eradicate such in order to create as near as possible a consistent near matt black background.

• 1^st – positioning the DPR card horizontally upon a clear white background permits DPR to see the DPR card in images taken, thereby compute the size of the stamp and using FRT (Facial Recognition Technology) to analyse the Perforation and calculate the perforation result.

• When we started testing DPR … we had similar unsatisfactory experiences to yours. Over time, persistent use and testing – we have refined not just DPR – but also identified best practice to get the best results out of the system.

Please bear with me …

Last night I was describing a perforate essay for TUT (TopUpTwenty) from Southern Rhodesia that the original collector owner had written up as P.14 – presumably because the final issued stamp was issued in P.14 in 1947… being somewhat tired, and the essay stamp perforation being a large stamp appearing somewhat different, as in larger – I took the opportunity to ‘DPR’ the perforation … and was surprised to learn that the stamp essay/trial was produced in P.12 ½ … when the final stamp was produced / issued P.14. I then used the ‘fiddly/challenging’ Instanta Perforation Gauge to verify my result. That confirmed P.12 ½ DPR reading doubled the value of asking price for the SPECIMEN colour trial.
STOP PRESS: this stamp sold online within 16 hours later! 19/12/24

A couple of other historical ‘pointers’ / story experiences come to mind:

1. In the 1980’s a new watermark ‘roller-system’ detector was introduced using an ink pouch to receive the impression of the watermark. It was called the ‘Morley-Bright’. For the life of me I never once got/took a satisfactory result – but Morley-Bright sold 1,000’s, I never heard an adverse comment … and a collector asked UPA if we could find one only a few months ago.
2. Then electronic watermark detectors came along – we have more than one – and we must use ONCE per annum – even though we check watermarks hundreds of times p/a.
   … so we don’t really get on with them though others may swear by them. Watermark Detectors cost up to £200= … my super-safe/clarity watermark fluid and watermark tray costs a fraction.
3. Perhaps 25 to 30 years ago the ‘Perfotronic 1’ and subsequent ‘Perfotronic 2’ materialised … at £300= each … (up to $600= US) … we have a couple – incredibly expensive and impractical with no internet connectivity and no ‘save’ / share functionality … (we have a couple – which we have not used once in 10 years other than recent testing against DPR).
4. Finally … we can use UV – Ultra-Violet lamps for stamp testing/checking … we get on with them – others/ many collectors don’t.

• Incidentally: I know of specialist dealers who have elected to create consistent examination conditions by having specific areas/light booths offering controlled conditions. This works for them …
  … come to think of it … I tend to do my DPR ‘perf-work’ under controlled conditions too.