Projective Homography Perspective Transformation and Local Binarization
Real-time computer vision processing in the client browser
When photographing paper documents with a mobile phone, perspective distortion occurs due to the camera angle. Our algorithm applies a 4-point projective homography transformation to map the tilted sheet onto a calibrated A4 aspect ratio.
An adaptive local thresholding filter (enhanced Otsu) separates dark text from background shadows, producing crisp black-and-white documents with small file sizes that look like they were produced by a commercial flatbed scanner.
Mobile Document Capture, Expense Invoices, and Urgent Submissions
Digitizing signed paper agreements, diplomas, and identity documents on the go without access to a flatbed scanner.
Submitting clear expense receipts and vendor invoices to accounting teams from your smartphone.
System Technical Specifications (スキャンしてPDF作成)
| Technical Parameter | System Specification |
|---|---|
| Vision Processing | 4-point projective homography transformation |
| Enhancement Filters | High-contrast document B&W, Grayscale, and Color Boost |
| Output Packaging | Standardized A4 PDF container with optimized DCTDecode compression |
| Supported Devices | Smartphones (iOS, Android), tablets, laptops with webcams |