问在DependencyObject上使用DependencyProperty.AddOwner方法而不是简单的POCO属性getter/setter的原因是什么？

EN

public class TestVisual: Shape
{
    private LineGeometry line = new LineGeometry();

    public static readonly DependencyProperty XY1Property =
        LineGeometry.StartPointProperty.AddOwner(
            typeof(TestVisual),
            new FrameworkPropertyMetadata(new Point(0,0),
                FrameworkPropertyMetadataOptions.AffectsMeasure));

    public static readonly DependencyProperty XY2Property =
        LineGeometry.EndPointProperty.AddOwner(
            typeof(TestVisual),
            new FrameworkPropertyMetadata(new Point(0, 0),
                FrameworkPropertyMetadataOptions.AffectsMeasure));


    public Point XY1
    {
        get { return (Point)GetValue(XY1Property);}
        set { SetValue(XY1Property,value); }
    }

    public Point XY2
    {
        get { return (Point)GetValue(XY2Property); }
        set { SetValue(XY2Property, value); }
    }


    protected override Geometry DefiningGeometry
    {
        get
        {
            line.StartPoint = XY1 ;
            line.EndPoint = XY2;                         
            return line;
        }
    }
}

    public class TestVisual: Shape
{
    private LineGeometry line = new LineGeometry();

    public Point XY1
    {
        get;
        set;
    }

    public Point XY2
    {
        get;
        set;
    }


    protected override Geometry DefiningGeometry
    {
        get
        {
            line.StartPoint = XY1 ;
            line.EndPoint = XY2;                         
            return line;
        }
    }
}