Adaptive transform coding incorporating time domain aliasing cancellation
Speech Communication
Discrete cosine transform filtering
Signal Processing
Discrete cosine transform: algorithms, advantages, applications
Discrete cosine transform: algorithms, advantages, applications
Discrete-time signal processing (2nd ed.)
Discrete-time signal processing (2nd ed.)
SIAM Review
Signal Processing with Lapped Transforms
Signal Processing with Lapped Transforms
Symmetric convolution and the discrete sine and cosine transforms
IEEE Transactions on Signal Processing
Fast DCT domain filtering using the DCT and the DST
IEEE Transactions on Image Processing
DFT spectrum estimation from critically sampled lapped transforms
Signal Processing
| Hi-index | 0.08 |
In this paper, expressions for convolution multiplication properties of DCT IV and DST IV are derived starting from equivalent DFT representations. Using these expressions methods for implementing linear filtering through block convolution in the DCT IV and DST IV domain are proposed. Techniques developed for DCT IV and DST IV are further extended to MDCT and MDST where the filter implementation is near exact for symmetric filters and approximate for non-symmetric filters. No additional overlapping is required for implementing the symmetric filtering in the MDCT domain and hence the proposed algorithm is computationally competitive with DFT based systems. Moreover, inherent 50% overlap between the adjacent frames used for MDCT/MDST domain reduces the blocking artifacts due to block processing or quantization. The techniques are computationally efficient for symmetric filters and provides a new alternative to DFT based convolution.