Image data compression method involving deleting data in areas where predicted color value, based on color change between adjacent pixels, is small, and image data processing device implementing same method

Abstract

An image data compression method in which reductions in data amount are achieved by deleting some pixel data. Compression is achieved by performing a reading step to read an image as plural pixel data and a deleting step to delete some of the pixel data. A prediction error calculating step calculates a prediction error that represents the difference between the color value pixel data of a pixel in question and the predicted color value thereof, as predicted by a prescribed prediction method from the pixel data of a pixel adjacent thereto. An area classifying step classifies the pixel data of the entire image area into either an abrupt color change areas type in which the prediction error is large or a nonabrupt color change area in which the prediction error is small. The deleting step is performed only on the pixel data classified in the nonabrupt color change area; deleted-position information is appended to the non-deleted pixel data that indicates the position of a pixel whose data has been deleted. This improves the quality of reproduction, since no pixel data are deleted where large errors would be expected to occur in reproduction.

Dependencies are inferred from the claim text.