Fractal image compression takes advantage of similarities within an image. It divides an image into small blocks and compares each block to larger parent blocks, storing the location and transform needed to match child and parent blocks. During decompression, it applies the transforms to reconstruct the image iteratively from a blank starting image. While it offers high theoretical compression, its greatest weakness is the time needed for encoding due to the large number of block comparisons required. Possible improvements include ordering blocks and limiting comparisons to reduce encoding time.