Color Image Compression via the Higher-Order Singular Value Decomposition (HOSVD)
Color Image Compression via the Higher-Order Singular Value Decomposition (HOSVD)
This Demonstration shows a method for data compression via the higher-order singular value decomposition (HOSVD) with color images interpreted as third-order tensors. The data compression is achieved by decreasing the tensor ranks in the pixel dimensions, while the third dimension (RGB channels) is not changed.
(x,y)
In the source code of this Demonstration, a basic general-purpose HOSVD toolbox is included as a collection of functions. It works with arbitrary tensor dimensions.