Clark School Home UMD
Return to Patents

DCT-Based Motion Estimation


A method for estimating the translational motion of an object is provided that first computes a plurality of DCT/DST coefficients for two sequential frames. These transform coefficients are then used to compute a pseudo phase function which is then fed into an inverse transform decoder. The output of the decoder is searched for peak values which are then used to determine the displacement of the object.


K. J. Ray Liu
Ut-Va Koc

Date Issued 


Patent No.