VIDEO COMPRESSION USING LAPPED TRANSFORMS FOR MOTION ESTIMATION COMPENSATION AND CODING/

Many conventional video coding schemes, such as the CCITT H.261 recomm endation, are based on the independent processing of nonoverlapping im age blocks. An important disadvantage with this approach is that block ing artifacts may be visible in the decoded frames. We propose a codin g scheme based entirely on the processing of overlapping, windowed dat a blocks, thus eliminating blocking effects. Motion estimation and, in part, compensation are performed in the frequency domain using a comp lex lapped transform (CLT), which can be viewed as a complex extension of the lapped orthogonal transform (LOT). The motion compensation alg orithm is equivalent to overlapped compensation in the spatial domain, but also allows image interpolation for subpixel displacements and so phisticated loop filters to be conveniently applied in the frequency d omain. For inter- and intraframe coding, we define the modified fast l apped transform (MFLT). This is a modified form of the LOT that entire ly eliminates blocking artifacts in the reconstructed data. The transf orm is applied in a hierarchical structure, and performs better than t he discrete cosine transform (DCT) for both coding modes. The proposed coder is compared with the H.261 scheme and is found to have signific antly improved performance.