feat: add table detection options and scan artifact removal

- Add TableDetectionSelector component for wired/wireless/region detection - Add CV-based table line detector module (disabled due to poor performance) - Add scan artifact removal preprocessing step (removes faint horizontal lines) - Add PreprocessingConfig schema with remove_scan_artifacts option - Update frontend PreprocessingSettings with scan artifact toggle - Integrate table detection config into ProcessingPage - Archive extract-table-cell-boxes proposal 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com>
2025-11-30 13:21:50 +08:00
parent f5a2c8a750
commit 95ae1f1bdb
17 changed files with 1906 additions and 344 deletions
--- a/frontend/src/components/PreprocessingSettings.tsx
+++ b/frontend/src/components/PreprocessingSettings.tsx
@@ -241,6 +241,25 @@ export default function PreprocessingSettings({
            )}
          </div>

+          {/* Scan Artifact Removal Toggle */}
+          <div className="space-y-2">
+            <label className="flex items-center gap-2 cursor-pointer">
+              <input
+                type="checkbox"
+                checked={config.remove_scan_artifacts}
+                onChange={(e) => handleConfigChange('remove_scan_artifacts', e.target.checked)}
+                disabled={disabled}
+                className="w-4 h-4 rounded border-gray-300 text-blue-600 focus:ring-blue-500"
+              />
+              <span className="text-sm text-gray-700">
+                {t('processing.preprocessing.removeScanArtifacts')}
+              </span>
+            </label>
+            <p className="text-xs text-gray-500 pl-6">
+              {t('processing.preprocessing.removeScanArtifactsDesc')}
+            </p>
+          </div>
+
          {/* Binarize Toggle - Hidden by default, shown only in advanced mode */}
          <details className="pt-2">
            <summary className="text-xs text-gray-500 cursor-pointer hover:text-gray-700">
--- a/frontend/src/components/TableDetectionSelector.tsx
+++ b/frontend/src/components/TableDetectionSelector.tsx
@@ -0,0 +1,124 @@
+import { cn } from '@/lib/utils'
+import { Checkbox } from '@/components/ui/checkbox'
+import { Table, Grid3X3, Rows3 } from 'lucide-react'
+import { useTranslation } from 'react-i18next'
+import type { TableDetectionConfig } from '@/types/apiV2'
+
+interface TableDetectionSelectorProps {
+  value: TableDetectionConfig
+  onChange: (config: TableDetectionConfig) => void
+  disabled?: boolean
+  className?: string
+}
+
+interface DetectionOption {
+  key: keyof TableDetectionConfig
+  icon: React.ReactNode
+  labelKey: string
+  descKey: string
+}
+
+const DETECTION_OPTIONS: DetectionOption[] = [
+  {
+    key: 'enable_wired_table',
+    icon: <Grid3X3 className="w-5 h-5" />,
+    labelKey: 'processing.tableDetection.wired',
+    descKey: 'processing.tableDetection.wiredDesc',
+  },
+  {
+    key: 'enable_wireless_table',
+    icon: <Rows3 className="w-5 h-5" />,
+    labelKey: 'processing.tableDetection.wireless',
+    descKey: 'processing.tableDetection.wirelessDesc',
+  },
+  {
+    key: 'enable_region_detection',
+    icon: <Table className="w-5 h-5" />,
+    labelKey: 'processing.tableDetection.region',
+    descKey: 'processing.tableDetection.regionDesc',
+  },
+]
+
+export default function TableDetectionSelector({
+  value,
+  onChange,
+  disabled = false,
+  className,
+}: TableDetectionSelectorProps) {
+  const { t } = useTranslation()
+
+  const handleOptionChange = (key: keyof TableDetectionConfig, checked: boolean) => {
+    onChange({
+      ...value,
+      [key]: checked,
+    })
+  }
+
+  return (
+    <div className={cn('border rounded-lg p-4 bg-white', className)}>
+      {/* Header */}
+      <div className="flex items-center gap-2 mb-4">
+        <Table className="w-5 h-5 text-gray-600" />
+        <h3 className="text-lg font-semibold text-gray-900">{t('processing.tableDetection.title')}</h3>
+      </div>
+
+      {/* Detection Options */}
+      <div className="space-y-3">
+        {DETECTION_OPTIONS.map((option) => {
+          const isChecked = value[option.key]
+
+          return (
+            <label
+              key={option.key}
+              className={cn(
+                'flex items-start gap-4 p-4 rounded-lg border-2 transition-all cursor-pointer',
+                isChecked
+                  ? 'border-blue-500 bg-blue-50'
+                  : 'border-gray-200 hover:border-gray-300 hover:bg-gray-50',
+                disabled && 'opacity-50 cursor-not-allowed'
+              )}
+            >
+              {/* Checkbox */}
+              <Checkbox
+                checked={isChecked}
+                onCheckedChange={(checked) => handleOptionChange(option.key, checked)}
+                disabled={disabled}
+                className="mt-0.5"
+              />
+
+              {/* Icon */}
+              <div
+                className={cn(
+                  'p-2 rounded-lg flex-shrink-0',
+                  isChecked ? 'bg-blue-100 text-blue-600' : 'bg-gray-100 text-gray-500'
+                )}
+              >
+                {option.icon}
+              </div>
+
+              {/* Content */}
+              <div className="flex-1 min-w-0">
+                <span
+                  className={cn(
+                    'font-medium',
+                    isChecked ? 'text-blue-700' : 'text-gray-900'
+                  )}
+                >
+                  {t(option.labelKey)}
+                </span>
+                <p className="text-sm text-gray-500 mt-1">{t(option.descKey)}</p>
+              </div>
+            </label>
+          )
+        })}
+      </div>
+
+      {/* Info Note */}
+      <div className="mt-4 p-3 bg-amber-50 border border-amber-200 rounded-md">
+        <p className="text-sm text-amber-800">
+          {t('processing.tableDetection.note')}
+        </p>
+      </div>
+    </div>
+  )
+}
--- a/frontend/src/i18n/locales/zh-TW.json
+++ b/frontend/src/i18n/locales/zh-TW.json
@@ -64,6 +64,16 @@
      "recommended": "推薦",
      "note": "版面模型會影響文件結構（表格、文字區塊、圖片）的偵測效果。請根據您的文件類型選擇適合的模型。"
    },
+    "tableDetection": {
+      "title": "表格偵測模式",
+      "wired": "有框線表格",
+      "wiredDesc": "偵測有明顯格線邊框的表格，適用於正式表格文件",
+      "wireless": "無框線表格",
+      "wirelessDesc": "偵測無邊框的表格，透過對齊方式推斷表格結構",
+      "region": "區域偵測",
+      "regionDesc": "輔助偵測表格區域，改善複雜表格的儲存格識別",
+      "note": "可同時啟用多種偵測模式，系統會自動整合偵測結果。如果表格儲存格框線不正確，請嘗試調整偵測模式。"
+    },
    "preprocessing": {
      "title": "影像前處理",
      "mode": {
@@ -92,6 +102,8 @@
        "strong": "強",
        "maximum": "最強"
      },
+      "removeScanArtifacts": "移除掃描瑕疵",
+      "removeScanArtifactsDesc": "移除掃描時光源產生的水平線痕，避免被誤判為表格框線",
      "advanced": "進階選項",
      "binarize": "二值化處理",
      "binarizeWarning": "不建議使用",
--- a/frontend/src/pages/ProcessingPage.tsx
+++ b/frontend/src/pages/ProcessingPage.tsx
@@ -12,9 +12,10 @@ import { Play, CheckCircle, FileText, AlertCircle, Clock, Activity, Loader2, Inf
 import LayoutModelSelector from '@/components/LayoutModelSelector'
 import PreprocessingSettings from '@/components/PreprocessingSettings'
 import PreprocessingPreview from '@/components/PreprocessingPreview'
+import TableDetectionSelector from '@/components/TableDetectionSelector'
 import TaskNotFound from '@/components/TaskNotFound'
 import { useTaskValidation } from '@/hooks/useTaskValidation'
-import type { LayoutModel, ProcessingOptions, PreprocessingMode, PreprocessingConfig, DocumentAnalysisResponse } from '@/types/apiV2'
+import type { LayoutModel, ProcessingOptions, PreprocessingMode, PreprocessingConfig, TableDetectionConfig, DocumentAnalysisResponse } from '@/types/apiV2'

 export default function ProcessingPage() {
  const { t } = useTranslation()
@@ -44,9 +45,17 @@ export default function ProcessingPage() {
    sharpen: true,
    sharpen_strength: 1.0,
    binarize: false,
+    remove_scan_artifacts: true,
  })
  const [showPreview, setShowPreview] = useState(false)

+  // Table detection state
+  const [tableDetectionConfig, setTableDetectionConfig] = useState<TableDetectionConfig>({
+    enable_wired_table: true,
+    enable_wireless_table: true,
+    enable_region_detection: true,
+  })
+
  // Analyze document to determine if OCR is needed (only for pending tasks)
  const { data: documentAnalysis, isLoading: isAnalyzing } = useQuery({
    queryKey: ['documentAnalysis', taskId],
@@ -70,6 +79,7 @@ export default function ProcessingPage() {
        layout_model: layoutModel,
        preprocessing_mode: preprocessingMode,
        preprocessing_config: preprocessingMode === 'manual' ? preprocessingConfig : undefined,
+        table_detection: tableDetectionConfig,
      }

      return apiClientV2.startTask(taskId!, options)
@@ -441,6 +451,13 @@ export default function ProcessingPage() {
                disabled={processOCRMutation.isPending}
              />

+              {/* Table Detection Settings */}
+              <TableDetectionSelector
+                value={tableDetectionConfig}
+                onChange={setTableDetectionConfig}
+                disabled={processOCRMutation.isPending}
+              />
+
              {/* Preprocessing Settings */}
              <PreprocessingSettings
                mode={preprocessingMode}
--- a/frontend/src/types/apiV2.ts
+++ b/frontend/src/types/apiV2.ts
@@ -108,6 +108,20 @@ export interface PreprocessingConfig {
  sharpen: boolean
  sharpen_strength: number   // 0.5-2.0, default 1.0
  binarize: boolean
+  remove_scan_artifacts: boolean  // Remove horizontal scan line artifacts
+}
+
+/**
+ * Table detection configuration for PP-StructureV3.
+ * Controls which table detection modes to enable.
+ * - enable_wired_table: Tables with visible cell borders/grid lines
+ * - enable_wireless_table: Tables without visible borders
+ * - enable_region_detection: Detect table-like regions for better cell structure
+ */
+export interface TableDetectionConfig {
+  enable_wired_table: boolean
+  enable_wireless_table: boolean
+  enable_region_detection: boolean
 }

 /**
@@ -147,6 +161,7 @@ export interface ProcessingOptions {
  layout_model?: LayoutModel  // Layout detection model selection (OCR track only)
  preprocessing_mode?: PreprocessingMode  // Preprocessing mode (OCR track only)
  preprocessing_config?: PreprocessingConfig  // Manual preprocessing config
+  table_detection?: TableDetectionConfig  // Table detection options (OCR track only)
 }

 export interface TaskCreate {