Image processing apparatus and control method thereof for converting multilevel image data to binary data using an error diffusion method

摘要

A technique for performing high-quality cluster type FM halftone image output without “anisotropy” or “sweep out texture” phenomenon is provided. An error diffused to a current pixel position and a random number RND are added by an adder to multi-level pixel data as a subject of binarization, and supplied as pixel data Pxy to a selector. The selector compares the pixel data Pxy with thresholds, and supplies the pixel data Pxy to one of a highlight portion processor, a dark portion processor and an intermediate portion processor. The highlight and dark portion processors refer to already-binarized results within a range determined in accordance with the value of the input pixel data Pxy, and determine a binarization result Qxy. The intermediate portion processor compares a threshold, which monotonously increases in accordance with the value of the pixel data Pxy, with the pixel data Pxy, thereby determines the binarization result Qxy.