Extract all text from this document image with formatting. Group elements into rows for DOCX table generation.

ROW GROUPING RULES:
1. Elements at SAME Y position (horizontal line) = same row
2. Elements that are VERTICALLY STACKED but form a logical block = same row
   - Example: "COC 9242925" stacked above "30 September 2023" on the right side → same row as title on left
   - These form a header block: [TITLE on left] [COC + Date stacked on right]
3. Multi-line paragraphs that form ONE continuous text = ONE element, ONE row

PARAGRAPH MERGING:
- If multiple lines form ONE logical paragraph (continuous sentence), merge into ONE element
- The bbox should span all merged lines
- Join text with spaces

SPECIAL ELEMENTS (MUST detect and output as placeholders):
- Photos/passport images: Output "[Photo affixed]" with element_type "photo_area" - ALWAYS detect photo boxes
- Signatures (handwritten): Output "[Signature affixed]" with element_type "signature_area"
- Stamps/seals (circular or rectangular): Output "[Stamp affixed]" with element_type "stamp_area"
- QR codes: Output "[QR Code]" with element_type "qr_code"
- Logos/emblems/crests/coat of arms: Output "[Logo]" with element_type "logo_area" - detect government emblems, company logos, official crests
- Numbered lists (1. 2. 3.): Keep number as "label", text as "value" in same row

IMPORTANT: Scan the ENTIRE document for non-text visual elements (logos, photos, stamps, QR codes) and include them!

FOOTER GROUPING:
- All footer elements (URL, date, page number) at bottom of page should be in ONE row

Output JSON format:
[
  {
    "row": 1,
    "elements": [
      {
        "bbox_2d": [x1, y1, x2, y2],
        "text_content": "...",
        "element_type": "title|header|paragraph|label|value|photo_area|signature_area|stamp_area|qr_code|logo_area|footer",
        "style": {
          "bold": true/false,
          "italic": true/false,
          "font_size": "small|normal|large|xlarge",
          "alignment": "left|center|right"
        }
      }
    ]
  }
]

FORMATTING DETECTION:
- bold: true if text appears bold/heavy
- italic: true if text is slanted
- font_size: small (<10pt), normal (10-12pt), large (14-18pt), xlarge (>18pt)
- alignment: left/center/right based on position in document