🔎 How Searchable PDF Maker Works
A Searchable PDF contains an invisible, selectable text layer positioned exactly behind the visual scan image. When searching with Ctrl+F in Adobe Reader, Chrome, or mobile viewers, the viewer searches and highlights this invisible text while preserving the authentic visual appearance of the original paper document.
This tool is specifically optimized for complex regional documents such as 3-column Bengali electoral voter rolls, land deeds (khatian/porcha), court decrees, and municipal records.
📑
Multi-Column Layout Slicing
Intelligently segments multi-column scanned tables into vertical slices to avoid reading order confusion across voter boxes.
🇧🇩
Bengali & English Dual OCR
Uses vision neural networks capable of recognizing Bengali conjunct characters, English names, and voter card alphanumeric serials.
👻
Zero-Opacity Invisible Text Layer
Injects text with 0% opacity (opacity: 0) aligned with original bounding box coordinates for seamless copy-pasting.
🤖 Secure Server-Side AI OCR Processing
Searchable PDF Maker runs on a secure Cloudflare Worker architecture with enterprise-grade protection:
- No Personal API Keys Required: You never need to obtain, paste, or manage API keys. Credentials are kept strictly secure within Cloudflare encrypted environment secrets.
- No Browser Credential Storage: Your browser never stores or handles any API tokens in
localStorage or sessionStorage.
- Encrypted Server-to-Server Processing: Requests communicate directly with our Cloudflare Worker endpoint (
/api/searchable-pdf-ai), which securely interfaces with AI vision models.
⚠️ OCR Accuracy Limitations & Quality Notes
While modern AI OCR is exceptionally capable, optical recognition is never 100% error-free:
- Scan Resolution: Clear 200–300 DPI scans yield the highest recognition accuracy. Blurred photos, distorted angles, or extreme shadows may lead to misread characters.
- Verification Recommended: Critical legal documents, voter serial numbers, and financial records should always be verified against the visual scan.
- Automatic Cooldown: When processing multi-page documents, the engine includes automatic cooldown pauses to respect API processing rate limits.
🔒 Privacy & Data Transmission Notice
Zero-Storage Architecture: Image slices are processed on-the-fly strictly in memory during OCR extraction. Pragati Telecom does not retain, store, or log your document contents. All buffers are automatically discarded once your PDF is compiled.