Method and device for single and/or multi-scale noise reduction system for pictures

Claim 3

A method as claimed in claim 1, wherein said sampling window function is a Gaussian or a binomial function, and said set of weighted polynomials are Hermite polynomials, or Krawtchouk polynomials, respectively.

Claim 4

A method as claimed in claim 1, wherein between steps (e) and (f) said method further comprises the steps of: aggregating a predetermined number of adjacent sampling window positions, as represented by their zeroth order polynomials, so as to form a larger sampling window; and performing steps (a) to (f) with respect to said larger sampling window.

Claim 5

A method as claimed in claim 3, wherein said sampling window function is a Gaussian or a binomial function, and said set of weighted polynomials are Hermite polynomials, or Krawtchouk polynomials, respectively.

Claim 6

A method as claimed in claim 3, wherein between steps (e) and (f) said method further comprises the steps of: aggregating a predetermined number of adjacent sampling window positions, as represented by their zeroth order polynomials, so as to form a still larger sampling window, and performing steps (a) to (f) with respect to said still larger sampling window.

Claim 7

A method as claimed in claim 5, wherein said sampling window function is a Gaussian or a binomial function, and said set of weighted polynomials are Hermite polynomials, or Krawtchouk polynomials, respectively.

Claim 8

Apparatus for reducing noise in a two-dimensional video image formed by pixels of various pixel intensities, comprising: windowing means for sampling the image with a uniform sampling window successively placed at each of a uniform array of sampling window positions, each pixel being included in one or more of the sampling window positions, the sampling deriving from each pixel a pixel value in each of the sampling window positions in which said pixel is included; multiplying means coupled to said windowing means for multiplying the pixel values in each sampling window position by a sampling window function so as to derive a set of multiplied pixel values for each sampling window position; forward transform means coupled to said multiplying means for performing a forward transform of the set of multiplied pixel values for each sampling window position, so as to derive for each sampling window position a finite set of weighted polynomials which are orthogonal to said sampling window function and include a zeroth order polynomial and polynomials of a first and further ascending orders, the respective orders of said set of weighted polynomials being produced at respective outputs of said forward transform means; intensity determining means coupled to a first order output of said forward transform means for generating from the first orders of said set of weighted polynomials for each sampling window position a scalar value (E 1 ) representing a spatial intensity variation across the relevant sampling window position; transmission control means coupled to the outputs of said forward transform means and to said intensity determining means and which, for each sampling window position, is adapted to detect whether said scalar value exceeds a preset threshold and to control transmission of the set of weighted polynomials at the outputs of said forward transform means so that (i) when said scalar value exceeds said threshold, all orders of said set of weighted polynomials are transmitted substantially equivalently, and (ii) when said scalar value does not exceed said threshold, predominantly the zeroth order of said set of polynomials are transmitted; and inverse transform means for receiving the set of weighted polynomials passed by said transmission control means for each sampling window position and performing thereon the inverse of said forward transform so as to derive a set of noise-reduced multiplied pixel values, and dividing the noise-reduced multiplied pixel values by said sampling window function so as to derive a set of noise-reduced pixel values representing a noise-reduced portion of said video image within the sampling window position.

Claim 9

Apparatus for reducing noise in a two-dimensional video image formed by pixels of various pixel intensities, comprising: windowing means for sampling the image with a uniform sampling window successively placed at each of a uniform array of sampling window positions, each pixel being included in one or more of the sampling window positions, the sampling deriving from each pixel a pixel value in each of the sampling window positions in which said pixel is included; multiplying means coupled to said windowing means for multiplying the pixel values in each sampling window position by a sampling window function so as to derive a set of multiplied pixel values for each sampling window position; forward transform means coupled to said multiplying means for performing a forward transform of the set of multiplied pixel values for each sampling window position, so as to derive for each sampling window position a finite set of weighted polynomials which are orthogonal to said sampling window function and include a zeroth order polynomial and polynomials of a first and further ascending orders, the respective orders of said set of weighted polynomials being produced at respective outputs of said forward transform means; intensity determining means coupled to a first order output of said forward transform means for generating from the first orders of said set of weighted polynomials for each sampling window position a scalar value (E 1 ) representing a spatial intensity variation across the relevant sampling window position; transmission control means coupled to the outputs of said forward transform means and to said intensity determining means and which, for each sampling window position, is adapted to detect whether said scalar value exceeds a preset threshold and to control transmission of the set of weighted polynomials at the outputs of said forward transform means so that (i) when said scalar value exceeds said threshold, all orders of said set of weighted polynomials are transmitted substantially equivalently, and (ii) when said scalar value does not exceed said threshold, predominantly the zeroth order of said set of polynomials are transmitted; and inverse transform means for receiving the set of weighted polynomials passed by said transmission control means for each sampling window position and performing thereon the inverse of said forward transform so as to derive a set of noise-reduced multiplied pixel values, and dividing the noise-reduced multiplied pixel values by said sampling window function so as to derive a set of noise-reduced pixel values representing a noise-reduced portion of said video image within the sampling window position.

Claim 10

An apparatus for reducing noise in a video image constituted by a two-dimensional array of pixels, said apparatus having successive levels each of which is an apparatus as claimed in claim 7, said levels being interconnected in pyramid relationship such that: the zeroth order output of said forward transform means in each successive level of said pyramid except the last level thereof is coupled to an input of the windowing means in the next higher level of said pyramid; an output of said inverse transform means in each successive level of said pyramid except the lowest level thereof is coupled to a zeroth order input of the inverse transform means in the next lower level to supply thereto pixel values in lieu of the lowest order of the set of polynomials produced by the forward transform means in said next lower level; and the sampling window employed in each successive level of said pyramid after the lowest level thereof is formed by a plurality of adjacent sampling windows employed in the immediately preceding level.