feat(ocr): Add CV Document Reconstruction Pipeline for vocabulary extraction
New OCR method using classical Computer Vision: high-res rendering (432 DPI), deskew, dewarp, binarization, projection-profile layout analysis, multi-pass Tesseract OCR with region-specific PSM, and Y-coordinate line alignment. Includes bugfix for convert_pdf_to_image call (line 869) and 39 unit tests. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
This commit is contained in:
BreakPilot Dev
@@ -93,6 +93,15 @@ const OCR_METHODS = {
|
||||
description: 'ARM64-nativ, Standard',
|
||||
enabled: true,
|
||||
},
|
||||
cv_pipeline: {
|
||||
id: 'cv_pipeline',
|
||||
name: 'Loesung E: Document Reconstruction',
|
||||
shortName: 'E: Doc Recon',
|
||||
model: 'opencv + tesseract (multi-pass)',
|
||||
color: 'green',
|
||||
description: 'CV-Pipeline: Deskew, Dewarp, Binarisierung, Multi-Pass OCR',
|
||||
enabled: true,
|
||||
},
|
||||
}
|
||||
|
||||
export default function OCRComparePage() {
|
||||
@@ -115,7 +124,7 @@ export default function OCRComparePage() {
|
||||
const [uploading, setUploading] = useState(false)
|
||||
|
||||
// Method Selection
|
||||
const [selectedMethods, setSelectedMethods] = useState<string[]>(['vision_llm', 'tesseract'])
|
||||
const [selectedMethods, setSelectedMethods] = useState<string[]>(['vision_llm', 'tesseract', 'cv_pipeline'])
|
||||
|
||||
// QR Upload State
|
||||
const [showQRModal, setShowQRModal] = useState(false)
|
||||
@@ -133,6 +142,9 @@ export default function OCRComparePage() {
|
||||
const [showGridOverlay, setShowGridOverlay] = useState(true)
|
||||
const [selectedCell, setSelectedCell] = useState<GridCell | null>(null)
|
||||
const [showCellDialog, setShowCellDialog] = useState(false)
|
||||
const [showMmGrid, setShowMmGrid] = useState(false)
|
||||
const [showTextAtPosition, setShowTextAtPosition] = useState(false)
|
||||
const [editableText, setEditableText] = useState(false)
|
||||
|
||||
// Block Review State
|
||||
const [blockReviewMode, setBlockReviewMode] = useState(false)
|
||||
@@ -651,6 +663,7 @@ export default function OCRComparePage() {
|
||||
blue: { bg: 'bg-blue-50', border: 'border-blue-300', text: 'text-blue-700' },
|
||||
red: { bg: 'bg-red-50', border: 'border-red-300', text: 'text-red-700' },
|
||||
purple: { bg: 'bg-purple-50', border: 'border-purple-300', text: 'text-purple-700' },
|
||||
green: { bg: 'bg-green-50', border: 'border-green-300', text: 'text-green-700' },
|
||||
}
|
||||
return colors[color]?.[type] || colors.slate[type]
|
||||
}
|
||||
@@ -839,7 +852,7 @@ export default function OCRComparePage() {
|
||||
{/* Grid Analysis Button */}
|
||||
<button
|
||||
onClick={analyzeGrid}
|
||||
disabled={analyzingGrid}
|
||||
disabled={analyzingGrid || !sessionId || !result}
|
||||
className="w-full px-4 py-2 bg-teal-600 text-white rounded-lg font-medium hover:bg-teal-700 disabled:opacity-50 disabled:cursor-not-allowed text-sm"
|
||||
>
|
||||
{analyzingGrid ? (
|
||||
@@ -874,6 +887,41 @@ export default function OCRComparePage() {
|
||||
<span className="text-slate-700">Grid-Overlay anzeigen</span>
|
||||
</label>
|
||||
|
||||
<label className="flex items-center gap-2 text-sm cursor-pointer">
|
||||
<input
|
||||
type="checkbox"
|
||||
checked={showMmGrid}
|
||||
onChange={(e) => setShowMmGrid(e.target.checked)}
|
||||
className="rounded"
|
||||
/>
|
||||
<span className="text-slate-700">1mm Raster anzeigen</span>
|
||||
</label>
|
||||
|
||||
<label className="flex items-center gap-2 text-sm cursor-pointer">
|
||||
<input
|
||||
type="checkbox"
|
||||
checked={showTextAtPosition}
|
||||
onChange={(e) => {
|
||||
setShowTextAtPosition(e.target.checked)
|
||||
if (!e.target.checked) setEditableText(false)
|
||||
}}
|
||||
className="rounded"
|
||||
/>
|
||||
<span className="text-slate-700">Text an Originalposition</span>
|
||||
</label>
|
||||
|
||||
{showTextAtPosition && (
|
||||
<label className="flex items-center gap-2 text-sm cursor-pointer ml-5">
|
||||
<input
|
||||
type="checkbox"
|
||||
checked={editableText}
|
||||
onChange={(e) => setEditableText(e.target.checked)}
|
||||
className="rounded"
|
||||
/>
|
||||
<span className="text-slate-700">Text bearbeitbar</span>
|
||||
</label>
|
||||
)}
|
||||
|
||||
{/* Block Review Button */}
|
||||
{result && nonEmptyBlockCount > 0 && (
|
||||
<button
|
||||
@@ -1122,7 +1170,20 @@ export default function OCRComparePage() {
|
||||
selectedCell={selectedCell}
|
||||
showEmpty={false}
|
||||
showNumbers={blockReviewMode}
|
||||
showTextLabels={true}
|
||||
showTextLabels={!showTextAtPosition}
|
||||
showMmGrid={showMmGrid}
|
||||
showTextAtPosition={showTextAtPosition}
|
||||
editableText={editableText}
|
||||
onCellTextChange={(cell, newText) => {
|
||||
if (!gridData) return
|
||||
const newCells = gridData.cells.map(row =>
|
||||
row.map(c => c.row === cell.row && c.col === cell.col
|
||||
? { ...c, text: newText, status: 'manual' as const }
|
||||
: c
|
||||
)
|
||||
)
|
||||
setGridData({ ...gridData, cells: newCells })
|
||||
}}
|
||||
highlightedBlockNumber={blockReviewMode ? currentBlockNumber : null}
|
||||
className={`rounded-lg border border-slate-200 overflow-hidden ${isFullscreen ? 'max-h-[80vh] mx-auto' : 'w-full max-w-2xl mx-auto'}`}
|
||||
/>
|
||||
@@ -1229,7 +1290,20 @@ export default function OCRComparePage() {
|
||||
selectedCell={selectedCell}
|
||||
showEmpty={false}
|
||||
showNumbers={blockReviewMode}
|
||||
showTextLabels={!blockReviewMode}
|
||||
showTextLabels={!blockReviewMode && !showTextAtPosition}
|
||||
showMmGrid={showMmGrid}
|
||||
showTextAtPosition={showTextAtPosition}
|
||||
editableText={editableText}
|
||||
onCellTextChange={(cell, newText) => {
|
||||
if (!gridData) return
|
||||
const newCells = gridData.cells.map(row =>
|
||||
row.map(c => c.row === cell.row && c.col === cell.col
|
||||
? { ...c, text: newText, status: 'manual' as const }
|
||||
: c
|
||||
)
|
||||
)
|
||||
setGridData({ ...gridData, cells: newCells })
|
||||
}}
|
||||
highlightedBlockNumber={blockReviewMode ? currentBlockNumber : null}
|
||||
className="rounded-lg border border-slate-200 overflow-hidden"
|
||||
/>
|
||||
|
||||
Reference in New Issue
Block a user