feat: add VIN photo OCR pipeline (refs #67)

Implement VIN-specific OCR extraction with optimized preprocessing:

- Add POST /extract/vin endpoint for VIN extraction
- VIN preprocessor: CLAHE, deskew, denoise, adaptive threshold
- VIN validator: check digit validation, OCR error correction (I->1, O->0)
- VIN extractor: PSM modes 6/7/8, character whitelist, alternatives
- Response includes confidence, bounding box, and alternatives
- Unit tests for validator and preprocessor
- Integration tests for VIN extraction endpoint

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

ocr/app/routers/extract.py

@@ -3,7 +3,8 @@ import logging
 
 from fastapi import APIRouter, File, HTTPException, Query, UploadFile
 
-from app.models import OcrResponse
+from app.extractors.vin_extractor import vin_extractor
+from app.models import BoundingBox, OcrResponse, VinAlternative, VinExtractionResponse
 from app.services import ocr_service
 
 logger = logging.getLogger(__name__)
@@ -67,3 +68,89 @@ async def extract_text(
         )
 
     return result
+
+
+@router.post("/vin", response_model=VinExtractionResponse)
+async def extract_vin(
+    file: UploadFile = File(..., description="Image file containing VIN"),
+) -> VinExtractionResponse:
+    """
+    Extract VIN (Vehicle Identification Number) from an uploaded image.
+
+    Uses VIN-optimized preprocessing and pattern matching:
+    - HEIC conversion (if needed)
+    - Grayscale conversion
+    - Deskew correction
+    - CLAHE contrast enhancement
+    - Noise reduction
+    - Adaptive thresholding
+    - VIN pattern matching (17 chars, excludes I/O/Q)
+    - Check digit validation
+    - Common OCR error correction (I->1, O->0, Q->0)
+
+    Supports HEIC, JPEG, PNG formats.
+    Processing time target: <3 seconds.
+
+    - **file**: Image file (max 10MB)
+
+    Returns:
+    - **vin**: Extracted VIN (17 alphanumeric characters)
+    - **confidence**: Confidence score (0.0-1.0)
+    - **boundingBox**: Location of VIN in image (if detected)
+    - **alternatives**: Other VIN candidates with confidence scores
+    - **processingTimeMs**: Processing time in milliseconds
+    """
+    # Validate file presence
+    if not file.filename:
+        raise HTTPException(status_code=400, detail="No file provided")
+
+    # Read file content
+    content = await file.read()
+    file_size = len(content)
+
+    # Validate file size
+    if file_size > MAX_SYNC_SIZE:
+        raise HTTPException(
+            status_code=413,
+            detail=f"File too large. Max: {MAX_SYNC_SIZE // (1024*1024)}MB",
+        )
+
+    if file_size == 0:
+        raise HTTPException(status_code=400, detail="Empty file provided")
+
+    logger.info(
+        f"VIN extraction: {file.filename}, "
+        f"size: {file_size} bytes, "
+        f"content_type: {file.content_type}"
+    )
+
+    # Perform VIN extraction
+    result = vin_extractor.extract(
+        image_bytes=content,
+        content_type=file.content_type,
+    )
+
+    # Convert internal result to API response
+    bounding_box = None
+    if result.bounding_box:
+        bounding_box = BoundingBox(
+            x=result.bounding_box.x,
+            y=result.bounding_box.y,
+            width=result.bounding_box.width,
+            height=result.bounding_box.height,
+        )
+
+    alternatives = [
+        VinAlternative(vin=alt.vin, confidence=alt.confidence)
+        for alt in result.alternatives
+    ]
+
+    return VinExtractionResponse(
+        success=result.success,
+        vin=result.vin,
+        confidence=result.confidence,
+        boundingBox=bounding_box,
+        alternatives=alternatives,
+        processingTimeMs=result.processing_time_ms,
+        error=result.error,
+    )