You can do this in two ways, both of which require 3D cameras and projection:
First, give all of your objects a 3D position in the world.
Next, create a camera with a position and orientation:
// Position of the camera in the world
// Place where the camera is looking
// Matrices representing the camera's parameters
You can update the camera's parameters like this:
// Creates a view matrix from where the camera is looking and its position.
ViewMatrix = Matrix.CreateLookAt(Position, Target, Vector3.Up);
// Creates a perspective matrix with 1.2 radian FOV, 16:9 aspect ratio,
// near plane of 0.01 units, and far plane of 10 units.
Matrix.CreatePerspectiveFOV(1.2f, 16.0f / 9.0f, 0.01f, 10.0f);
Now, the two ways of drawing.
SpriteBatch, the Hacky Way
You're probably familiar with SpriteBatch already. It's a way of drawing sprites to the screen in screen space. One thing you can do is simply project all of your objects onto the screen, scale them by their distance to the camera, and then render them. This is a hacky way of doing things, because you have to fake perspective and the depth buffer.
Here's how to do it:
// Draw each entity in the scene
foreach(Entity thing in Scene)
// Project its position onto the screen using the camera.
= GraphicsDevice.Viewport.Project(thing.Position, camera.ProjectionMatrix,
// Determine if its on the screen.
bool isOnScreen = Viewport.Bounds.Contains((int)screenPos.X, (int)screenPos.Y);
// Compute the depth and scale of the object.
float depth = screenPos.Z;
float scale = 1.0f / depth;
// Draw it at the appropriate depth and scale.
SpriteBatch.Draw(thing.Image, new Vector2(screenPos.X, screenPos.Y),
thing.SourceRect, thing.Tint, thing.Rotation, thing.Origin,
new Vector2(scale, scale), thing.SpriteEffects, depth);
Textured Billboards, the Correct Way
You can also just create for each sprite a physical 3D model, just a 3D billboard, oriented to face the camera. Then, you can render it like any other XNA 3D geometry. The depth buffer will work flawlessly, you won't have to guess the scale, and perspective will be taken care of correctly. I suggest this tutorial to get you started.