The imagepoints are kind of irrelevant when it comes to the memory. The only concerning memory will be the main map image, if it is a huge HD image then it will take up more memory, and best practice is to use smaller source images. If this is just going to be one image and not say 20 huge maps at a time then it shouldn't creep up too much. The invisible detection sprites will also be kind of irrelevant, you won't be using any images, just an invisible sprite object for detection. The only other memory use will come from the highlighting sprite which will be represented by an image of just the one state, this shouldn't be too much memory and is likely not going to be detailed as it is just for highlighting a selected area.