OCR and searchable PDFs for court filing: why scans bounce

It is late, the draft is settled, and the annexures are scans — photographs of pages taken on a phone, prints run twice through the chamber copier, attachments a client emailed in 2019. They open. They print. On screen they look like documents. To an e-filing system, most of them are pictures of documents — and that difference is where filings run into objections.

A scan is a photograph. The registry wants a document.

Optical character recognition — OCR, since the term now appears on scrutiny lists — is the step that places a layer of machine-readable text underneath a scanned page. The page itself does not change: the image the court sees is the image you sent, stamp, signature and all. What changes is what sits beneath it. With the text layer in place, the PDF can be searched, its passages copied, an annexure found by a word instead of a scroll.

Without it, the file is an album. Press Ctrl+F, type a word you can plainly see on the page, and the search returns nothing — because as far as the software is concerned, there are no words there.

Why image-only scans draw objections

Three reasons, all mundane, all curable.

The rules ask for it in terms. Most High Court e-filing rules — Delhi's among them — require uploaded pleadings and annexures to be OCR-searchable, and registry scrutiny checks for it the way it checks pagination. "Not OCR searchable" is a standard entry on objection lists, alongside its cousins: pages not legible, font below the prescribed size, page numbers wrong or missing.

The registry's hierarchy of sins is unforgiving at the bottom. Page numbers are the one thing no registry overlooks; legibility comes close behind. A bad scan can fail on both at once — a photographed page that clipped the number, a faint page nobody can read.

A searchable record is a usable record. A bookmarked, OCR'd paperbook lets a reader land on annexure C-14 by its title, or find every mention of a clause in one search. An image-only bundle offers the scroll wheel.

What a properly built filing PDF carries

Every paperbook AIC returns carries four things as a matter of course:

• A text layer under every page — the whole record OCR'd, front to back.
• Bookmarks — each annexure bookmarked under the exact title your draft gave it, never a title of ours.
• Continuous pagination — one run of page numbers across the whole set, volumes included.
• A master index — every annexure mapped to its page, in your pleaded order.

None of this touches your document. The text sits under the image; the titles are your titles; the order is your order. We flag; we never edit.

The same work also stands on its own. Not every job is a full paperbook: send a single PDF and it comes back OCR-searchable and bookmarked to its own index — the same discipline, none of the assembly.

The page OCR cannot save

OCR reads what the eye can read. A dim page, a faded carbon, a part-scanned drawing — the software will not invent what is not there, and neither will AIC: anything faint or cropped is flagged for a human eye. We flag; we never guess.

The cure for a genuinely illegible page is older than the software: a faithful, court-standard typed copy, set into the same page and the same pagination — because 31 is just 31, and a 31A and 31B squeezed in after the fact is its own objection. We type and confirm; certifying the true copy stays with you, as it must.

Files are read-only, processed on our own machines, deleted on completion. The clerk, not counsel.

OCR and e-filing: questions litigators ask

How do I check whether my PDF is searchable?

Open it, press Ctrl+F (Cmd+F on a Mac) and search for a word you can see on a scanned page. If the search finds nothing, that page has no text layer — it is an image, whatever the file extension says.

Does OCR alter my document?

No. The text layer sits underneath the scanned image; the page the court reads is exactly the page you sent. Your titles, your order, your words — reproduced, never rephrased. We flag; we never edit.

Can OCR fix a dim or faded page?

No. OCR reads what the eye can read, and flags what it cannot. The recognised cure is a court-standard typed copy set into the same page and pagination, which AIC prepares and confirms — certifying it as a true copy remains counsel's act.

Can I send one PDF for OCR and bookmarking, without a full paperbook?

Yes. OCR and bookmarking stands on its own: send the PDF as it sits, and it comes back searchable and bookmarked to its own index — read-only, processed on our own machines, deleted on completion.