The three-dimensional error diffusion propagates errors not only into the current frame, but also into the next frame. Therefore, the error filter of the three-dimensional error diffusion is more symmetric than the error filters of the traditional met...
The three-dimensional error diffusion propagates errors not only into the current frame, but also into the next frame. Therefore, the error filter of the three-dimensional error diffusion is more symmetric than the error filters of the traditional methods. However, the weights of the three-dimensional filter show greater fluctuations according to the amount of boundary information. Because of this, images have strong high-frequency components, and the accumulated error in a specific location causes unwanted dot patterns. The proposed method does not use edge information. This method changes the error propagation direction into the time dimension and presents suitable filter coefficients. The error filter of this method propagates diffusion error into the time dimension according to motion information between two frames. Motion information is gained through a level difference between current pixels of two frames. Through experiments, we confirmed that the proposed method can lead to the smooth gradation and effectively reduce the stop patterns in the dynamic image.