An image processing apparatus includes a dividing unit configured to divide a target candidate image, which is assigned as a candidate of a target image, into blocks when a photo-mosaic image is generated by dividing the target image into the blocks and replacing each block with a raw image selected from multiple raw images; a block-class classification unit configured to place each block in the target candidate image into a class selected from multiple classes defined by a characteristic associated with color; a comparing unit configured to compare the number of blocks in the target candidate image in a class and the number of raw images in the same class; and a selecting unit configured to select a target candidate image among a multiple target candidate images to be assigned as the target image on the basis of comparison results of the target candidate images obtained by the comparing unit.