A scanned invoice or receipt photo is an image, such as JPEG, PNG, TIFF or HEIC. The pictured invoice is pixels rather than named invoice fields. Text recognition can read those pixels, but the date, amounts and VAT still need checking, especially with blur or faded paper. It is not a structured e-invoice. Klarbon support is planned, not available yet.
What is in an image file?
| Format | Typical source | Notes |
|---|---|---|
| JPEG | phone cameras, scanners | lossy compression, ITU-T T.81 |
| PNG | screenshots | lossless compression, W3C Recommendation |
| HEIC | newer phone cameras | HEIF photo format; compression and size depend on the image and settings |
| TIFF | office scanners | can hold several pages |
| Part | Holds |
|---|---|
| Pixels | the picture of the paper, at the scan or camera resolution |
| Metadata | date taken, device, sometimes location (EXIF) |
| Structured invoice fields | not provided by the receipt picture; metadata or OCR text may exist separately |
The W3C publishes the PNG specification, and ITU-T publishes JPEG as T.81 (PNG, JPEG).
Receipt image or e-invoice?
| Receipt image | EN 16931 e-invoice | |
|---|---|---|
| Holds | a picture of paper | named fields with codes |
| Reading it | text recognition, then a person's check | read named fields, then check the values |
| Can be checked against the rules | no | yes |
A PDF invoice without XML that was made from a scan is also an image.
What does Klarbon do with receipt images?
Klarbon does not open images today; reading invoice and receipt details from a photo or scan, with a result you confirm, is planned, not available yet. Klarbon creates no document archive. To read an e-invoice, use the invoice viewer.
Related
Frequently asked questions
Is a photo of an invoice an e-invoice?
No. A photo or scan is a picture of a paper invoice. It has no structured data, so it is not an e-invoice, even when it is sent by email.
Which image formats are used for receipts?
Phones save photos as JPEG or HEIC, screenshots are often PNG, and office scanners produce JPEG, TIFF or PDF. A scan saved as PDF is still an image inside the PDF.
How is text read from a receipt image?
With optical character recognition (OCR), which turns the pixels into text, and then a step that finds the date, amounts, VAT and seller in that text. Blur, folds, faded thermal paper and handwriting cause mistakes, so a person should confirm the result.
What makes a receipt photo easier to read?
Flat paper, even light without shadows, the whole receipt in the frame and the camera held straight above it. Crop out the background before saving.
Sources
- W3C, PNG specification source, living draft raw.githubusercontent.com
- Apple, HEIF and HEVC media on Apple devices support.apple.com