To correct the variation in the output characteristics of a plurality of output elements by the small amount of data as compared with conventional art with a high precision.

An image output device 10 includes a plurality of recording elements 3 for outputting an image for correction, a flat bed scanner 70 for acquiring read information from the image for the correction, and a correction processing unit 60 which obtains the amount of corrections of output density by each recording element 3 based on read information. The correction processing unit 60 calculates a second correction value c(i) for correcting an approximate function obtained under the condition that an output density y(x, i) becomes an average reference density yAverage(xo, I) to a (logx-logz') table when an input signal x is a reference input signal xo, from virtual input/output function when for each recording element 3 from a first correction value H(i) for correcting the reference light quantity value z(xo, i) corresponding to the average reference density yAverage (xo, I) to the reference input signal xo and virtual input/output function when corrected by the first correction value H(i) for each recording element 3, and corrects the input signal with these correction values.