Non-monotonic contour diffusion and algorithm
AssigneeTexas Instruments Incorporated
InventorsTodd A. Clatanoff, Stephen W. Marshall, Vishal Markandey
Life of this patent
Estimating the term…
Abstract
A method and device for producing an accurate approximation of a digital input word translated by a monotonic transfer function. The digital word is translated into a non-monotonic output word comprised of a closest-approximation component and an error component. The error component is zero for regions in which each consecutive input word produces a unique closest-approximation component. In regions in which each consecutive input word produces the same closest-approximation component, the error signal represents the number of consecutive input words which produce the same closest-approximation component, and the position within that run of consecutive input words occupied by the present input word.
Claims11
A semiconductor memory lookup table comprising:
- J memory locations, each said memory location corresponding to an input data word in a range of M consecutive input words;
- output words stored in each of said J memory locations, said output words each comprised of N binary bits representing the sum of:
- a first component representing a function of said range of input words, wherein contiguous sub-ranges of said range of input words translate to an identical N-bit binary output word;
- and
- a second component representing an error between said first component and a desired value.
A semiconductor memory lookup table comprising:
- J memory locations, each said memory location corresponding to an input data word in a range of M input words;
- outputs words stored in each of said J memory locations, said output words having a value representative of a function of said range of input words, wherein contiguous sub-ranges of said range of M input words translate to a common output word, at least a first of said sub-ranges abutting a second of said sub-ranges at a transition point, wherein said value of at least one said output word in said first of said sub-ranges is exchanged with said value of at least one said output word in said second of said sub-ranges.
Dependencies are inferred from the claim text.