Sprite collisions or bounding box?

So I'm wondering in terms of performance which is better. Lets say I have 100 buildings, 50 trees, 40 stream, and 40 mountain sprites all with solid sections. Would it improve performance to create 1invisible solid sprite and overlay copies of it all over the solid sections of those sprites. Or set collisions up for each of those sprites individually? I've been going with the assumption that using just the one sprite with collisions for all of my images would improve performance and it does seem to flicker less when scrolling the visible window across the layout. But is it really improving the logic or am I just imagining it?

Ya, less sprites the better. Use the display sprite's collision polygon if you can.