In this paper, we propose an inverse halftoning method based on Invertible Neural Network (INN), using grayscale images and halftone images as paired inputs for training. The proposed method yields enhanced quality for inverse halftone images in comparison to previous work, as demonstrated through measurements of peak signal-to-noise ratio (PSNR) and structural similarity index (SSIM) between the resultant inverse halftone image and the original continuous-tone image. Furthermore, the INN's inherent capacity enables us to accurately reconstruct the original halftone image from the inverse halftone image.