Abstract: A robust halftone image watermarking method was proposed. The method was developed in parity domain based on pixel block. Especially, the parity sum of a pixel block was defined by comparing the average of the pixel block with an image-dependent threshold. By altering the pixel block’s parity based on noise-balanced block error diffusion, watermark was spread into the host image. Watermark was retrieved by employing each pixel block’s parity and majority voting strategy, not referring to the original image. Compared with the state-of-the-art method in parity domain, the results indicate that the proposed method has high watermark rate and watermark rate flexibility. Moreover, it is capable of extracting watermark directly from the attacked watermarked image without quantizing it into a halftone image. And it can achieve high robustness against common attacks and print-and-scan attack of different types of printers and scanners.