Multiplier-free implementation of DCT used in image and video processing and compression
AssigneeHewlett-Packard Company
InventorsNeri Merhav, Vasudev Bhaskaran
Life of this patent
Estimating the term…
Abstract
Multiplier-free implementation of an approximation of the DCT used in image and video processing. In accordance with the primary aspect of the present invention, image and video processing is done with no multiplications and a fewer number of operations through the application of a modified Arai, Agui, and Nakajima (AAN) scheme for eight-point DCT.
Claims19
In a system that compresses or decompresses image and video pixel data using a discrete cosine transform process, an improvement comprising:
- factorizing the transform process into a number of matrices;
- quantizing at least one of the matrices, to thereby produce at least one corresponding quantized matrix having values selected from the set of numbers including zero, and positive and negative numbers that are powers of two;
- adjusting the values of at least one matrix that was not quantized to thereby compensate for errors which would otherwise occur as a result of quantizing values, to thereby produce at least one adjusted matrix;
- and
- providing hardware logic that performs the transform process upon input data, including operations derived from each quantized matrix and each adjusted matrix.
In a system that compresses or decompresses image and video pixel data using a discrete cosine transform process to process input data, a method comprising:
- factorizing the transform process into a number of matrices;
- quantizing at least one matrix to thereby obtain at least one quantized matrix having values selected from the set of numbers including zero, and positive and negative numbers that are powers of two;
- adjusting values of at least one matrix that has not been quantized, to thereby reduce errors that would otherwise be introduced by quantizing to thereby obtain at least one adjusted matrix;
- performing the transform process by substantially hardware-only processing of the input data, where the hardware implements operations dependent upon each quantized matrix and each adjusted matrix.
In a system that compresses or decompresses image and video pixel data using a discrete cosine transform process to process input data, a method comprising:
- factorizing the transform process into a number of matrices;
- quantizing at least one matrix to thereby obtain at least one quantized matrix having values selected from the set of numbers including zero and positive and negative numbers that are powers of two;
- adjusting values of at least one matrix that has not been quantized, to thereby reduce errors that would otherwise be introduced by quantizing to thereby obtain at least one adjusted matrix;
- performing the transform process upon the input data using operations dependent upon each quantized matrix and each adjusted matrix.
Dependencies are inferred from the claim text.