Content deleted Content added
No edit summary |
No edit summary |
||
Line 47:
Since the algorithm converges slowly, and, due to limitations in numerical precision the algorithm will often not converge, real-world applications of Lloyd's algorithm stop once the distribution is "good enough." One common termination criterion is when the maximum distance a point moves in one iteration is below some set limit.
The Lloyd's method was originally used for scalar quantization, but it is clear that the method extends for vector quantization as well. As such, it is extensively used in [[data compression]] techniques in [[information theory]]. Lloyd's method is used in computer graphics because the resulting distribution has [[blue noise]] characteristics (see also [[Colors of noise]]), meaning there are few low-frequency components that could be interpreted as artifacts. It is particularly well-suited to picking sample positions for [[dithering]].
Lloyd's algorithm is also used to generate dot drawings in the style of [[stippling]] {{harv|Deussen|Hiller|van Overveld|Strothotte|2000}}. In this application, the centroids can be weighted based on a reference image {{harv|Secord|2002}} to produce stipple illustrations matching an input image.
| |||