Method and device to enhance image quality in digital video processing systems using dithering

摘要

A processing chain for a digital image signal (12) applies a dither pattern (14), having a first spectrum, to the image signal at a point in the processing chain. A further noise pattern (10) is applied to the image signal during the processing chain. The noise pattern (10) has a second spectrum which is configured such that the combination of the first spectrum and second spectrum results in a more continuous spectrum. Another aspect describes a noise pattern (10) which can be used as an offset dither pattern for digital images, especially before color bit depth reduction. The noise pattern comprises an array of values which are linearly distributed across a range, with each value in the range occurring an equal number of times. Similar values at extreme ends of the range of values are dispersed within the array. The pattern has a Poisson-disk two-dimensional spectral energy distribution. Values are positioned in the array based on distance to similar values in neighboring repetitions of the array. The array has “magic square” properties.

主权项

1. A method of processing a digital image signal comprising
applying a noise pattern to the image signal, the noise pattern comprising an array of offset values, the values of the array being arranged in columns and lines and the array of offset values having a range of values and a spectral energy distribution, wherein the noise pattern has the following properties: the values in the array are linearly distributed across the range of values and each value in the range of values occurs a substantially equal number of times within the array; values close to or at extreme ends of the range of values are dispersed from each other within the array; the spectral energy distribution is a substantially Poisson-disk two-dimensional spectral energy distribution; and wherein the noise pattern also has at least one of the following further properties: values are positioned in the array based on distance to values that are close in value to the respective values in neighbouring repetitions of the array; a sum of said values in any column of the array is substantially equal to a sum of said values in any other column of the array; and a sum of said values in any line of the array is substantially equal to a sum of said values in any other line of the array.