In embodiment 1, a description will be given of an image processing apparatus that estimates a printing result to be printed based on a modeling-target printer and print medium with an estimation unit using two neural networks. Note that in the present disclosure, a printing result is an image printed on a print medium.

Generally, an image, i.e., a printing result, printed on a print medium by a printer has random scattering or bleeding of a color material, such as toner or ink, unique to the type of the printer and the type of the print medium. Thus, edges in the printing result meander and blur due to the random scattering or bleeding of the color material. In a case where a single neural network is used to simulate scanned image data of such a printing result, estimated image data with a strong blur than that of the actual scanned image data will be outputted.

To solve this, in the present embodiment, a printing result is estimated with two different estimation units connected in series. The estimation unit at the first stage outputs intermediate data being image data indicating regions on a print medium surface where a color material is applied at a predetermined density or higher, and reproducing the scattering or bleeding of the color material. The estimation unit at the second stage outputs estimated image data representing a printing result estimated from the intermediate data outputted from the estimation unit at the first stage, which the intermediate data reproduces the gradations, i.e., blurs, around edges resulting from the change in the density of the color material. In this manner, estimated image data representing a printing result is outputted with higher accuracy and at higher speed.

<Configuration of Image Processing Apparatus>