Spatially Compensated Block-based Transform

T.K. Tan (Singapore) and C.S. Boon (Japan)


Transform Coding, Image Compression, Video Compression, Adaptive Transform, Directional Transform.


Spatial transforms are commonly used in today’s image and video compression algorithms. Spatial transformation de-correlates the signal and concentrates the energy in low frequency coefficients prior to quantization and entropy coding processes. A method for performing spatial transformation where information adjacent to the main axis of the transform direction is used to further reduce the energy of the high band signal is investigated. The transform is implemented using a lifting method comprising of a prediction and an update step. These steps are augmented with a filtering operation that displaces the signal to be subtracted or added by a spatial shift vector. This results in improve energy compaction compared to traditional transforms, while maintaining the separable and orthogonal properties of the transform.

Important Links:

Go Back