将 Windows Ink 笔划识别为文本和形状

项目
04/06/2023

使用 Windows Ink 内置的识别功能将笔划墨迹转换为文本和形状。

重要 API：InkCanvas、Windows.UI.Input.Inking

带有墨迹分析的自由格式识别

我们在此处演示如何使用 Windows Ink 分析引擎 (Windows.UI.Input.Inking.Analysis 将 InkCanvas 上的一组自由格式笔划分类、分析和识别为文本或形状。（除了文本和形状识别外，墨迹分析还可用于识别文档结构、项目符号列表和通用绘画。）

注意

对于基本的单行纯文本情形（如表单输入），请参阅本主题后面的受限制的手写识别。

在本示例中，用户在单击某个按钮以指示完成绘画时，将会启动识别。

从墨迹分析示例（基本）下载此示例

首先，我们设置 UI (MainPage.xaml)。

UI 包含一个“识别”按钮、一个 InkCanvas 和一个标准 Canvas。当按下“识别”按钮时，将会分析墨迹画布上的所有墨迹笔划，并将对应的形状和文本绘制在标准画布上（如果成功识别）。原始的笔划墨迹随后将从墨迹画布中删除。

<Grid Background="{ThemeResource ApplicationPageBackgroundThemeBrush}">
     <Grid.RowDefinitions>
         <RowDefinition Height="Auto"/>
         <RowDefinition Height="*"/>
     </Grid.RowDefinitions>
     <StackPanel x:Name="HeaderPanel" 
                 Orientation="Horizontal" 
                 Grid.Row="0">
         <TextBlock x:Name="Header" 
                     Text="Basic ink analysis sample" 
                     Style="{ThemeResource HeaderTextBlockStyle}" 
                     Margin="10,0,0,0" />
         <Button x:Name="recognize" 
                 Content="Recognize" 
                 Margin="50,0,10,0"/>
     </StackPanel>
     <Grid x:Name="drawingCanvas" Grid.Row="1">

         <!-- The canvas where we render the replacement text and shapes. -->
         <Canvas x:Name="recognitionCanvas" />
         <!-- The canvas for ink input. -->
         <InkCanvas x:Name="inkCanvas" />

     </Grid>
</Grid>

在 UI 代码隐藏文件 (MainPage.xaml.cs) 中，添加墨迹和墨迹分析功能所需的命名空间类型引用：

然后，我们指定全局变量：

 InkAnalyzer inkAnalyzer = new InkAnalyzer();
 IReadOnlyList<InkStroke> inkStrokes = null;
 InkAnalysisResult inkAnalysisResults = null;

接着，我们设置一些基本墨迹输入行为：

将 InkPresenter 配置为将来自笔、鼠标和触摸的输入数据解释为墨迹笔划 (InputDeviceTypes)。
笔划墨迹使用指定的 InkDrawingAttributes 在 InkCanvas 上呈现。
还声明一个用于“识别”按钮上的单击事件的侦听器。

/// <summary>
/// Initialize the UI page.
/// </summary>
public MainPage()
{
    this.InitializeComponent();

    // Set supported inking device types.
    inkCanvas.InkPresenter.InputDeviceTypes =
        Windows.UI.Core.CoreInputDeviceTypes.Mouse |
        Windows.UI.Core.CoreInputDeviceTypes.Pen | 
        Windows.UI.Core.CoreInputDeviceTypes.Touch;

    // Set initial ink stroke attributes.
    InkDrawingAttributes drawingAttributes = new InkDrawingAttributes();
    drawingAttributes.Color = Windows.UI.Colors.Black;
    drawingAttributes.IgnorePressure = false;
    drawingAttributes.FitToCurve = true;
    inkCanvas.InkPresenter.UpdateDefaultDrawingAttributes(drawingAttributes);

    // Listen for button click to initiate recognition.
    recognize.Click += RecognizeStrokes_Click;
}

在本例中，我们在“识别”按钮的单击事件处理程序中执行墨迹分析。

首先，调用 InkCanvas.InkPresenter的 StrokeContainer 上的 GetStrokes，以获得当前所有墨迹笔划的集合。
如果存在墨迹笔划，则通过调用将其传递至 InkAnalyzer 的 AddDataForStrokes。
我们正在尝试识别绘画和文本，但你可以使用 SetStrokeDataKind 方法指定是只对文本（包括文档结构和项目符号列表）感兴趣，还是只对绘画（包括形状识别）感兴趣。
调用 AnalyzeAsync 以启动墨迹分析并获得 InkAnalysisResult。
如果 Status 返回已更新状态，请调用 InkAnalysisNodeKind.InkWord 和 InkAnalysisNodeKind.InkDrawing 的 FindNodes。
重复两组节点类型，并在识别画布上（墨迹画布下面）绘制各自的文本或形状。
最后，删除 InkAnalyzer 中已识别的节点和墨迹画布中对应的墨迹笔划。

/// <summary>
/// The "Analyze" button click handler.
/// Ink recognition is performed here.
/// </summary>
/// <param name="sender">Source of the click event</param>
/// <param name="e">Event args for the button click routed event</param>
private async void RecognizeStrokes_Click(object sender, RoutedEventArgs e)
{
    inkStrokes = inkCanvas.InkPresenter.StrokeContainer.GetStrokes();
    // Ensure an ink stroke is present.
    if (inkStrokes.Count > 0)
    {
        inkAnalyzer.AddDataForStrokes(inkStrokes);

        // In this example, we try to recognizing both 
        // writing and drawing, so the platform default 
        // of "InkAnalysisStrokeKind.Auto" is used.
        // If you're only interested in a specific type of recognition,
        // such as writing or drawing, you can constrain recognition 
        // using the SetStrokDataKind method as follows:
        // foreach (var stroke in strokesText)
        // {
        //     analyzerText.SetStrokeDataKind(
        //      stroke.Id, InkAnalysisStrokeKind.Writing);
        // }
        // This can improve both efficiency and recognition results.
        inkAnalysisResults = await inkAnalyzer.AnalyzeAsync();

        // Have ink strokes on the canvas changed?
        if (inkAnalysisResults.Status == InkAnalysisStatus.Updated)
        {
            // Find all strokes that are recognized as handwriting and 
            // create a corresponding ink analysis InkWord node.
            var inkwordNodes = 
                inkAnalyzer.AnalysisRoot.FindNodes(
                    InkAnalysisNodeKind.InkWord);

            // Iterate through each InkWord node.
            // Draw primary recognized text on recognitionCanvas 
            // (for this example, we ignore alternatives), and delete 
            // ink analysis data and recognized strokes.
            foreach (InkAnalysisInkWord node in inkwordNodes)
            {
                // Draw a TextBlock object on the recognitionCanvas.
                DrawText(node.RecognizedText, node.BoundingRect);

                foreach (var strokeId in node.GetStrokeIds())
                {
                    var stroke = 
                        inkCanvas.InkPresenter.StrokeContainer.GetStrokeById(strokeId);
                    stroke.Selected = true;
                }
                inkAnalyzer.RemoveDataForStrokes(node.GetStrokeIds());
            }
            inkCanvas.InkPresenter.StrokeContainer.DeleteSelected();

            // Find all strokes that are recognized as a drawing and 
            // create a corresponding ink analysis InkDrawing node.
            var inkdrawingNodes =
                inkAnalyzer.AnalysisRoot.FindNodes(
                    InkAnalysisNodeKind.InkDrawing);
            // Iterate through each InkDrawing node.
            // Draw recognized shapes on recognitionCanvas and
            // delete ink analysis data and recognized strokes.
            foreach (InkAnalysisInkDrawing node in inkdrawingNodes)
            {
                if (node.DrawingKind == InkAnalysisDrawingKind.Drawing)
                {
                    // Catch and process unsupported shapes (lines and so on) here.
                }
                // Process generalized shapes here (ellipses and polygons).
                else
                {
                    // Draw an Ellipse object on the recognitionCanvas (circle is a specialized ellipse).
                    if (node.DrawingKind == InkAnalysisDrawingKind.Circle || node.DrawingKind == InkAnalysisDrawingKind.Ellipse)
                    {
                        DrawEllipse(node);
                    }
                    // Draw a Polygon object on the recognitionCanvas.
                    else
                    {
                        DrawPolygon(node);
                    }
                    foreach (var strokeId in node.GetStrokeIds())
                    {
                        var stroke = inkCanvas.InkPresenter.StrokeContainer.GetStrokeById(strokeId);
                        stroke.Selected = true;
                    }
                }
                inkAnalyzer.RemoveDataForStrokes(node.GetStrokeIds());
            }
            inkCanvas.InkPresenter.StrokeContainer.DeleteSelected();
        }
    }
}

以下是在我们的识别画布上绘制 TextBlock 的函数。我们在墨迹画布上使用关联墨迹笔划的边界矩形来设置 TextBlock 的位置和字体大小。

 /// <summary>
 /// Draw ink recognition text string on the recognitionCanvas.
 /// </summary>
 /// <param name="recognizedText">The string returned by text recognition.</param>
 /// <param name="boundingRect">The bounding rect of the original ink writing.</param>
 private void DrawText(string recognizedText, Rect boundingRect)
 {
     TextBlock text = new TextBlock();
     Canvas.SetTop(text, boundingRect.Top);
     Canvas.SetLeft(text, boundingRect.Left);

     text.Text = recognizedText;
     text.FontSize = boundingRect.Height;

     recognitionCanvas.Children.Add(text);
 }

以下是在我们的识别画布上绘制椭圆和多边形的函数。我们在墨迹画布上使用关联墨迹笔划的边界矩形来设置形状的位置和字体大小。

 // Draw an ellipse on the recognitionCanvas.
 private void DrawEllipse(InkAnalysisInkDrawing shape)
 {
     var points = shape.Points;
     Ellipse ellipse = new Ellipse();

     ellipse.Width = shape.BoundingRect.Width;
     ellipse.Height = shape.BoundingRect.Height;

     Canvas.SetTop(ellipse, shape.BoundingRect.Top);
     Canvas.SetLeft(ellipse, shape.BoundingRect.Left);

     var brush = new SolidColorBrush(Windows.UI.ColorHelper.FromArgb(255, 0, 0, 255));
     ellipse.Stroke = brush;
     ellipse.StrokeThickness = 2;
     recognitionCanvas.Children.Add(ellipse);
 }

 // Draw a polygon on the recognitionCanvas.
 private void DrawPolygon(InkAnalysisInkDrawing shape)
 {
     List<Point> points = new List<Point>(shape.Points);
     Polygon polygon = new Polygon();

     foreach (Point point in points)
     {
         polygon.Points.Add(point);
     }

     var brush = new SolidColorBrush(Windows.UI.ColorHelper.FromArgb(255, 0, 0, 255));
     polygon.Stroke = brush;
     polygon.StrokeThickness = 2;
     recognitionCanvas.Children.Add(polygon);
 }

以下是正在操作的示例：

分析前	分析后

受限制的手写识别

在上一个部分（带有墨迹分析的自由格式识别）中，我们演示了如何使用墨迹分析 API 分析和识别 InkCanvas 区域内的任意墨迹笔划。

在本部分中，我们将演示如何使用 Windows Ink 手写识别引擎（不是墨迹分析）将 InkCanvas 上的一组笔划转换为文本（基于已安装的默认语言包）。

注意

本部分所示的基本手写识别最适合于单行文本输入情形，如表单输入。如果是包含文档结构、列表项、形状和绘画的分析和解释的更丰富识别情形（包括文本识别在内），请参阅上一部分：带有墨迹分析的自由格式识别。

在本示例中，用户在单击某个按钮以指示完成书写时，将会启动识别。

从墨迹手写识别示例下载此示例

首先，我们设置 UI。

此 UI 包括一个“识别”按钮、InkCanvas 以及一个用于显示识别结果的区域。

<Grid Background="{ThemeResource ApplicationPageBackgroundThemeBrush}">
     <Grid.RowDefinitions>
         <RowDefinition Height="Auto"/>
         <RowDefinition Height="*"/>
     </Grid.RowDefinitions>
     <StackPanel x:Name="HeaderPanel"
                 Orientation="Horizontal"
                 Grid.Row="0">
             <TextBlock x:Name="Header"
                 Text="Basic ink recognition sample"
                 Style="{ThemeResource HeaderTextBlockStyle}"
                 Margin="10,0,0,0" />
             <Button x:Name="recognize"
                 Content="Recognize"
                 Margin="50,0,10,0"/>
     </StackPanel>
     <Grid Grid.Row="1">
         <Grid.RowDefinitions>
             <RowDefinition Height="*"/>
             <RowDefinition Height="Auto"/>
         </Grid.RowDefinitions>
         <InkCanvas x:Name="inkCanvas"
             Grid.Row="0"/>
         <TextBlock x:Name="recognitionResult"
             Grid.Row="1"
             Margin="50,0,10,0"/>
     </Grid>
 </Grid>

对于本示例，你需要首先添加墨迹功能所需的命名空间类型引用：
- Windows.UI.Input
- Windows.UI.Input.Inking

然后，我们设置一些基本墨迹输入行为。

将 InkPresenter 配置为将来自笔和鼠标的输入数据解释为笔划墨迹 (InputDeviceTypes)。笔划墨迹使用指定的 InkDrawingAttributes 在 InkCanvas 上呈现。还声明一个用于“识别”按钮上的单击事件的侦听器。

public MainPage()
    {
        this.InitializeComponent();

        // Set supported inking device types.
        inkCanvas.InkPresenter.InputDeviceTypes =
            Windows.UI.Core.CoreInputDeviceTypes.Mouse |
            Windows.UI.Core.CoreInputDeviceTypes.Pen;

        // Set initial ink stroke attributes.
        InkDrawingAttributes drawingAttributes = new InkDrawingAttributes();
        drawingAttributes.Color = Windows.UI.Colors.Black;
        drawingAttributes.IgnorePressure = false;
        drawingAttributes.FitToCurve = true;
        inkCanvas.InkPresenter.UpdateDefaultDrawingAttributes(drawingAttributes);

        // Listen for button click to initiate recognition.
        recognize.Click += Recognize_Click;
    }

最后，执行基本手写识别。在本例中，我们使用“识别”按钮的单击事件处理程序来执行手写识别。

InkPresenter 将所有笔划墨迹存储在 InkStrokeContainer 对象中。笔划通过 InkPresenter 的 StrokeContainer 属性公开，并使用 GetStrokes 方法检索。

// Get all strokes on the InkCanvas.
    IReadOnlyList<InkStroke> currentStrokes = inkCanvas.InkPresenter.StrokeContainer.GetStrokes();

创建一个 InkRecognizerContainer 来管理手写识别过程。

// Create a manager for the InkRecognizer object
    // used in handwriting recognition.
    InkRecognizerContainer inkRecognizerContainer =
        new InkRecognizerContainer();

调用 RecognizeAsync 以检索一组 InkRecognitionResult 对象。为 InkRecognizer 所检测到的每个单词生成识别结果。

// Recognize all ink strokes on the ink canvas.
    IReadOnlyList<InkRecognitionResult> recognitionResults =
        await inkRecognizerContainer.RecognizeAsync(
            inkCanvas.InkPresenter.StrokeContainer,
            InkRecognitionTarget.All);

每个 InkRecognitionResult 对象都包含一组文本候选项。识别引擎将此列表中最顶层的项视为最佳匹配项，后跟其余候选项（按置信度大小进行降序排序）。

我们循环访问每个 InkRecognitionResult 并编译候选项列表。然后显示候选项，并清除 InkStrokeContainer（也清除 InkCanvas）。

string str = "Recognition result\n";
    // Iterate through the recognition results.
    foreach (var result in recognitionResults)
    {
        // Get all recognition candidates from each recognition result.
        IReadOnlyList<string> candidates = result.GetTextCandidates();
        str += "Candidates: " + candidates.Count.ToString() + "\n";
        foreach (string candidate in candidates)
        {
            str += candidate + " ";
        }
    }
    // Display the recognition candidates.
    recognitionResult.Text = str;
    // Clear the ink canvas once recognition is complete.
    inkCanvas.InkPresenter.StrokeContainer.Clear();

下面是完整的点击事件处理程序示例。

// Handle button click to initiate recognition.
    private async void Recognize_Click(object sender, RoutedEventArgs e)
    {
        // Get all strokes on the InkCanvas.
        IReadOnlyList<InkStroke> currentStrokes = inkCanvas.InkPresenter.StrokeContainer.GetStrokes();

        // Ensure an ink stroke is present.
        if (currentStrokes.Count > 0)
        {
            // Create a manager for the InkRecognizer object
            // used in handwriting recognition.
            InkRecognizerContainer inkRecognizerContainer =
                new InkRecognizerContainer();

            // inkRecognizerContainer is null if a recognition engine is not available.
            if (!(inkRecognizerContainer == null))
            {
                // Recognize all ink strokes on the ink canvas.
                IReadOnlyList<InkRecognitionResult> recognitionResults =
                    await inkRecognizerContainer.RecognizeAsync(
                        inkCanvas.InkPresenter.StrokeContainer,
                        InkRecognitionTarget.All);
                // Process and display the recognition results.
                if (recognitionResults.Count > 0)
                {
                    string str = "Recognition result\n";
                    // Iterate through the recognition results.
                    foreach (var result in recognitionResults)
                    {
                        // Get all recognition candidates from each recognition result.
                        IReadOnlyList<string> candidates = result.GetTextCandidates();
                        str += "Candidates: " + candidates.Count.ToString() + "\n";
                        foreach (string candidate in candidates)
                        {
                            str += candidate + " ";
                        }
                    }
                    // Display the recognition candidates.
                    recognitionResult.Text = str;
                    // Clear the ink canvas once recognition is complete.
                    inkCanvas.InkPresenter.StrokeContainer.Clear();
                }
                else
                {
                    recognitionResult.Text = "No recognition results.";
                }
            }
            else
            {
                Windows.UI.Popups.MessageDialog messageDialog = new Windows.UI.Popups.MessageDialog("You must install handwriting recognition engine.");
                await messageDialog.ShowAsync();
            }
        }
        else
        {
            recognitionResult.Text = "No ink strokes to recognize.";
        }
    }

国际识别

内置于 Windows 墨迹平台的手写识别包括 Windows 支持的一小组广泛的区域设置和语言。

有关 InkRecognizer 支持的语言列表，请参阅 InkRecognizer.Name 属性主题。

你的应用可以查询已安装的手写识别引擎的集合，并使用其中一个引擎或让用户选择其首选语言。

注意用户可以通过转到“设置->时间和语言”来查看已安装语言的列表。已安装的语言在“语言”下列出。

若要安装新语言包并为该语言启用手写识别：

转到“设置”>“时间和语言”>“区域和语言”。
选择“添加语言”。
从列表中选择某种语言，然后选择区域版本。该语言现在在“区域和语言”页面上列出。
单击该语言，然后选择“选项”。
在“语言选项”页面上，下载“手写识别引擎”（也可以在此处下载完整的语言包、语音识别引擎和键盘布局）。

我们在此处演示如何使用手写识别引擎基于所选的识别器来解释 InkCanvas 上的一组笔划。

用户通过在完成书写时单击某个按钮来启动识别。

首先，我们设置 UI。

UI 包含一个“识别”按钮、一个列出已安装手写识别器的组合框、InkCanvas 和一个用于显示识别结果的区域。

<Grid Background="{ThemeResource ApplicationPageBackgroundThemeBrush}">
        <Grid.RowDefinitions>
            <RowDefinition Height="Auto"/>
            <RowDefinition Height="*"/>
        </Grid.RowDefinitions>
        <StackPanel x:Name="HeaderPanel"
                    Orientation="Horizontal"
                    Grid.Row="0">
            <TextBlock x:Name="Header"
                       Text="Advanced international ink recognition sample"
                       Style="{ThemeResource HeaderTextBlockStyle}"
                       Margin="10,0,0,0" />
            <ComboBox x:Name="comboInstalledRecognizers"
                     Margin="50,0,10,0">
                <ComboBox.ItemTemplate>
                    <DataTemplate>
                        <StackPanel Orientation="Horizontal">
                            <TextBlock Text="{Binding Name}" />
                        </StackPanel>
                    </DataTemplate>
                </ComboBox.ItemTemplate>
            </ComboBox>
            <Button x:Name="buttonRecognize"
                    Content="Recognize"
                    IsEnabled="False"
                    Margin="50,0,10,0"/>
        </StackPanel>
        <Grid Grid.Row="1">
            <Grid.RowDefinitions>
                <RowDefinition Height="*"/>
                <RowDefinition Height="Auto"/>
            </Grid.RowDefinitions>
            <InkCanvas x:Name="inkCanvas"
                       Grid.Row="0"/>
            <TextBlock x:Name="recognitionResult"
                       Grid.Row="1"
                       Margin="50,0,10,0"/>
        </Grid>
    </Grid>

然后，我们设置一些基本墨迹输入行为。

将 InkPresenter 配置为将来自笔和鼠标的输入数据解释为笔划墨迹 (InputDeviceTypes)。笔划墨迹使用指定的 InkDrawingAttributes 在 InkCanvas 上呈现。

我们调用 InitializeRecognizerList 函数以使用已安装手写识别器的列表来填充识别器组合框。

我们还声明“识别”按钮上的单击事件的侦听器和识别组合框上的选择更改事件。

 public MainPage()
 {
    this.InitializeComponent();

    // Set supported inking device types.
    inkCanvas.InkPresenter.InputDeviceTypes =
        Windows.UI.Core.CoreInputDeviceTypes.Mouse |
        Windows.UI.Core.CoreInputDeviceTypes.Pen;

    // Set initial ink stroke attributes.
    InkDrawingAttributes drawingAttributes = new InkDrawingAttributes();
    drawingAttributes.Color = Windows.UI.Colors.Black;
    drawingAttributes.IgnorePressure = false;
    drawingAttributes.FitToCurve = true;
    inkCanvas.InkPresenter.UpdateDefaultDrawingAttributes(drawingAttributes);

    // Populate the recognizer combo box with installed recognizers.
    InitializeRecognizerList();

    // Listen for combo box selection.
    comboInstalledRecognizers.SelectionChanged +=
        comboInstalledRecognizers_SelectionChanged;

    // Listen for button click to initiate recognition.
    buttonRecognize.Click += Recognize_Click;
}

我们使用已安装手写识别器的列表来填充识别器组合框。

创建一个 InkRecognizerContainer 来管理手写识别过程。使用此对象调用 GetRecognizers 并检索已安装识别器的列表来填充识别器组合框。

// Populate the recognizer combo box with installed recognizers.
private void InitializeRecognizerList()
{
    // Create a manager for the handwriting recognition process.
    inkRecognizerContainer = new InkRecognizerContainer();
    // Retrieve the collection of installed handwriting recognizers.
    IReadOnlyList<InkRecognizer> installedRecognizers =
        inkRecognizerContainer.GetRecognizers();
    // inkRecognizerContainer is null if a recognition engine is not available.
    if (!(inkRecognizerContainer == null))
    {
        comboInstalledRecognizers.ItemsSource = installedRecognizers;
        buttonRecognize.IsEnabled = true;
    }
}

如果识别器组合框选择发生更改，请更新手写识别器。

基于识别器组合框中所选的识别器使用 InkRecognizerContainer 调用 SetDefaultRecognizer。

// Handle recognizer change.
    private void comboInstalledRecognizers_SelectionChanged(
        object sender, SelectionChangedEventArgs e)
    {
        inkRecognizerContainer.SetDefaultRecognizer(
            (InkRecognizer)comboInstalledRecognizers.SelectedItem);
    }

最后，我们基于所选的手写识别器执行手写识别。在本例中，我们使用“识别”按钮的单击事件处理程序来执行手写识别。

InkPresenter 将所有笔划墨迹存储在 InkStrokeContainer 对象中。笔划通过 InkPresenter 的 StrokeContainer 属性公开，并使用 GetStrokes 方法检索。

// Get all strokes on the InkCanvas.
    IReadOnlyList<InkStroke> currentStrokes =
        inkCanvas.InkPresenter.StrokeContainer.GetStrokes();

调用 RecognizeAsync 以检索一组 InkRecognitionResult 对象。

为 InkRecognizer 所检测到的每个单词生成识别结果。

// Recognize all ink strokes on the ink canvas.
IReadOnlyList<InkRecognitionResult> recognitionResults =
    await inkRecognizerContainer.RecognizeAsync(
        inkCanvas.InkPresenter.StrokeContainer,
        InkRecognitionTarget.All);

每个 InkRecognitionResult 对象都包含一组文本候选项。识别引擎将此列表中最顶层的项视为最佳匹配项，后跟其余候选项（按置信度大小进行降序排序）。

我们循环访问每个 InkRecognitionResult 并编译候选项列表。然后显示候选项，并清除 InkStrokeContainer（也清除 InkCanvas）。

string str = "Recognition result\n";
// Iterate through the recognition results.
foreach (InkRecognitionResult result in recognitionResults)
{
        // Get all recognition candidates from each recognition result.
        IReadOnlyList<string> candidates =
            result.GetTextCandidates();
        str += "Candidates: " + candidates.Count.ToString() + "\n";
        foreach (string candidate in candidates)
        {
            str += candidate + " ";
        }
}
// Display the recognition candidates.
recognitionResult.Text = str;
// Clear the ink canvas once recognition is complete.
inkCanvas.InkPresenter.StrokeContainer.Clear();

下面是完整的点击事件处理程序示例。

// Handle button click to initiate recognition.
private async void Recognize_Click(object sender, RoutedEventArgs e)
{
    // Get all strokes on the InkCanvas.
    IReadOnlyList<InkStroke> currentStrokes =
        inkCanvas.InkPresenter.StrokeContainer.GetStrokes();

    // Ensure an ink stroke is present.
    if (currentStrokes.Count > 0)
    {
        // inkRecognizerContainer is null if a recognition engine is not available.
        if (!(inkRecognizerContainer == null))
        {
            // Recognize all ink strokes on the ink canvas.
            IReadOnlyList<InkRecognitionResult> recognitionResults =
                await inkRecognizerContainer.RecognizeAsync(
                    inkCanvas.InkPresenter.StrokeContainer,
                    InkRecognitionTarget.All);
            // Process and display the recognition results.
            if (recognitionResults.Count > 0)
            {
                string str = "Recognition result\n";
                // Iterate through the recognition results.
                foreach (InkRecognitionResult result in recognitionResults)
                {
                    // Get all recognition candidates from each recognition result.
                    IReadOnlyList<string> candidates =
                        result.GetTextCandidates();
                    str += "Candidates: " + candidates.Count.ToString() + "\n";
                    foreach (string candidate in candidates)
                    {
                        str += candidate + " ";
                    }
                }
                // Display the recognition candidates.
                recognitionResult.Text = str;
                // Clear the ink canvas once recognition is complete.
                inkCanvas.InkPresenter.StrokeContainer.Clear();
            }
            else
            {
                recognitionResult.Text = "No recognition results.";
            }
        }
        else
        {
            Windows.UI.Popups.MessageDialog messageDialog =
                new Windows.UI.Popups.MessageDialog(
                    "You must install handwriting recognition engine.");
            await messageDialog.ShowAsync();
        }
    }
    else
    {
        recognitionResult.Text = "No ink strokes to recognize.";
    }
}

动态识别

虽然前面的两个示例要求用户按下按钮来开始识别，但是你也可以使用与基本计时函数配对的笔划输入来执行动态识别。

在此示例中，我们将使用与前面的国际识别示例相同的 UI 和笔划设置。

这些全球对象（InkAnalyzer、InkStroke、InkAnalysisResult、DispatcherTimer）用于整个应用。
```
// Stroke recognition globals.
InkAnalyzer inkAnalyzer;
DispatcherTimer recoTimer;
```

我们将添加两个 InkPresenter 笔划事件（StrokesCollected 和 StrokeStarted）的侦听器（而不是用于启动识别的按钮），并设置一个带有一秒 Tick 间隔的基本计时器 (DispatcherTimer)。

public MainPage()
{
    this.InitializeComponent();

    // Set supported inking device types.
    inkCanvas.InkPresenter.InputDeviceTypes =
        Windows.UI.Core.CoreInputDeviceTypes.Mouse |
        Windows.UI.Core.CoreInputDeviceTypes.Pen;

    // Listen for stroke events on the InkPresenter to 
    // enable dynamic recognition.
    // StrokesCollected is fired when the user stops inking by 
    // lifting their pen or finger, or releasing the mouse button.
    inkCanvas.InkPresenter.StrokesCollected += inkCanvas_StrokesCollected;
    // StrokeStarted is fired when ink input is first detected.
    inkCanvas.InkPresenter.StrokeInput.StrokeStarted +=
        inkCanvas_StrokeStarted;

    inkAnalyzer = new InkAnalyzer();

    // Timer to manage dynamic recognition.
    recoTimer = new DispatcherTimer();
    recoTimer.Interval = TimeSpan.FromSeconds(1);
    recoTimer.Tick += recoTimer_TickAsync;
}

然后，我们为第一步中声明的 InkPresenter 事件定义处理程序（我们还替代 OnNavigatingFrom 页面事件以管理我们的计时器）。

StrokesCollected
将墨迹笔划 (AddDataForStrokes) 添加到 InkAnalyzer，当用户（通过抬起笔或手指或释放鼠标按钮）停止墨迹书写时，启动识别计时器。在无墨迹输入的一秒后，启动识别。

使用 SetStrokeDataKind 方法指定是只对文本（包括文档结构和项目符号列表）感兴趣，还是只对绘画（包括形状识别）感兴趣。
StrokeStarted
如果新的笔划在下一个计时器滴答事件前开始，则停止计时器，因为新的笔划可能是单次手写输入的延续。

// Handler for the InkPresenter StrokeStarted event.
// Don't perform analysis while a stroke is in progress.
// If a new stroke starts before the next timer tick event,
// stop the timer as the new stroke is likely the continuation
// of a single handwriting entry.
private void inkCanvas_StrokeStarted(InkStrokeInput sender, PointerEventArgs args)
{
    recoTimer.Stop();
}
// Handler for the InkPresenter StrokesCollected event.
// Stop the timer and add the collected strokes to the InkAnalyzer.
// Start the recognition timer when the user stops inking (by 
// lifting their pen or finger, or releasing the mouse button).
// If ink input is not detected after one second, initiate recognition.
private void inkCanvas_StrokesCollected(InkPresenter sender, InkStrokesCollectedEventArgs args)
{
    recoTimer.Stop();
    // If you're only interested in a specific type of recognition,
    // such as writing or drawing, you can constrain recognition 
    // using the SetStrokDataKind method, which can improve both 
    // efficiency and recognition results.
    // In this example, "InkAnalysisStrokeKind.Writing" is used.
    foreach (var stroke in args.Strokes)
    {
        inkAnalyzer.AddDataForStroke(stroke);
        inkAnalyzer.SetStrokeDataKind(stroke.Id, InkAnalysisStrokeKind.Writing);
    }
    recoTimer.Start();
}
// Override the Page OnNavigatingFrom event handler to 
// stop our timer if user leaves page.
protected override void OnNavigatingFrom(NavigatingCancelEventArgs e)
{
    recoTimer.Stop();
}

最后，执行手写识别。在此示例中，我们使用 DispatcherTimer 的 Tick 事件处理程序来启动手写识别。

调用 AnalyzeAsync 以启动墨迹分析并获得 InkAnalysisResult。
如果 Status 返回“已更新”状态，请为 InkAnalysisNodeKind.InkWord 节点类型调用 FindNodes。
重复节点并显示识别出的文本。
最后，删除 InkAnalyzer 中已识别的节点和墨迹画布中对应的墨迹笔划。

private async void recoTimer_TickAsync(object sender, object e)
{
    recoTimer.Stop();
    if (!inkAnalyzer.IsAnalyzing)
    {
        InkAnalysisResult result = await inkAnalyzer.AnalyzeAsync();

        // Have ink strokes on the canvas changed?
        if (result.Status == InkAnalysisStatus.Updated)
        {
            // Find all strokes that are recognized as handwriting and 
            // create a corresponding ink analysis InkWord node.
            var inkwordNodes =
                inkAnalyzer.AnalysisRoot.FindNodes(
                    InkAnalysisNodeKind.InkWord);

            // Iterate through each InkWord node.
            // Display the primary recognized text (for this example, 
            // we ignore alternatives), and then delete the 
            // ink analysis data and recognized strokes.
            foreach (InkAnalysisInkWord node in inkwordNodes)
            {
                string recognizedText = node.RecognizedText;
                // Display the recognition candidates.
                recognitionResult.Text = recognizedText;

                foreach (var strokeId in node.GetStrokeIds())
                {
                    var stroke =
                        inkCanvas.InkPresenter.StrokeContainer.GetStrokeById(strokeId);
                    stroke.Selected = true;
                }
                inkAnalyzer.RemoveDataForStrokes(node.GetStrokeIds());
            }
            inkCanvas.InkPresenter.StrokeContainer.DeleteSelected();
        }
    }
    else
    {
        // Ink analyzer is busy. Wait a while and try again.
        recoTimer.Start();
    }
}

笔和触笔交互

通过

将 Windows Ink 笔划识别为文本和形状

带有墨迹分析的自由格式识别

受限制的手写识别

国际识别

动态识别

主题示例

其他示例

其他资源

通过

将 Windows Ink 笔划识别为文本和形状

带有墨迹分析的自由格式识别

受限制的手写识别

国际识别

动态识别

相关文章

主题示例

其他示例

其他资源