The way I solved this problem was to place the image within a Border with it's ClipToBounds property set to True. The RenderTransformOrigin on the image is then set to 0.5,0.5 so the image will start zooming on the center of the image. The RenderTransform is also set to a TransformGroup containing a ScaleTransform and a TranslateTransform.
I then handled the MouseWheel event on the image to implement zooming
private void image_MouseWheel(object sender, MouseWheelEventArgs e)
{
var st = (ScaleTransform)image.RenderTransform;
double zoom = e.Delta > 0 ? .2 : -.2;
st.ScaleX += zoom;
st.ScaleY += zoom;
}
To handle the panning the first thing I did was to handle the MouseLeftButtonDown event on the image, to capture the mouse and to record it's location, I also store the current value of the TranslateTransform, this what is updated to implement panning.
Point start;
Point origin;
private void image_MouseLeftButtonDown(object sender, MouseButtonEventArgs e)
{
image.CaptureMouse();
var tt = (TranslateTransform)((TransformGroup)image.RenderTransform)
.Children.First(tr => tr is TranslateTransform);
start = e.GetPosition(border);
origin = new Point(tt.X, tt.Y);
}
Then I handled the MouseMove event to update the TranslateTransform.
private void image_MouseMove(object sender, MouseEventArgs e)
{
if (image.IsMouseCaptured)
{
var tt = (TranslateTransform)((TransformGroup)image.RenderTransform)
.Children.First(tr => tr is TranslateTransform);
Vector v = start - e.GetPosition(border);
tt.X = origin.X - v.X;
tt.Y = origin.Y - v.Y;
}
}
Finally don't forget to release the mouse capture.
private void image_MouseLeftButtonUp(object sender, MouseButtonEventArgs e)
{
image.ReleaseMouseCapture();
}
As for the selection handles for resizing this can be accomplished using an adorner, check out this article for more information.