An image processing device may generate text image data and background image data, based on original image data. The background image data may be generated by performing a correction process on target image data for setting a value of a first-type pixel having a tone value within a specific range before the correction process is performed to a tone value within the specific range after the correction process is performed, and setting a value of a second-type pixel having a tone value out of the specific range before the correction process is performed to a tone value within the specific range after the correction process is performed. The specific range may include a plurality of consecutive tone values that include a peak tone value. The peak tone value is included in a tone value distribution of the original image data and corresponds to a background of a text.