An image processing apparatus is capable of determining the optimal fixing condition
taking the local toner distribution (toner density) into consideration from image
data. The toner amount is obtained for a local area formed by a marked pixel and
the surrounding pixels per color component signal, and the toner amount of each
local area is added. Further, over the image data entirely, the toner amount of
local area is worked out in plural local areas. Then, a toner amount that indicates
the maximum value is acquired from the toner amount of plural local areas. On the
bases of the toner amount in the local area that indicates the maximum value and
the kind of sheet on which toner is fixed, the fixing speed is controlled optimally.