| 摘要 |
A source image containing both relatively low frequency information (i.e. graphics, pictures and background) and relatively high frequency information (i.e. text) are input to an image processing unit. This image processing unit can be embodied in functional units of an output device such as an LCD projector or may be part of a personal computer connected to the output device, or may even be part of an input device that provides the source image. The source image will be in one resolution and the output device will have a different, generally lower, resolution. The source image is segmented into a black text image and a white text image (i.e. portions of the source image formed as image blocks or tiles). The black text image and the white text image are then subtracted from the source image to form a background image. Then, the background image, the black text image and the white text image are filtered separately using different filters for each. The filtered or down-sampled images are then merged. This merged image can then be projection enlarged, in the example wherein the output device is an LCD projector, or printed or displayed in the examples wherein the output device is a printer or CRT, respectively. The filter for the background image is preferably a gaussian filter and the filters for the black text image and the white text images are preferably a combination of a gaussian filter (having different filter coefficients than the filter for the background image) and a sigmoid filter.
|
