Data processor

Abstract

Image data of a pixel matrix are subjected to orthogonal transformation to coefficient data, which are then quantized with a first quantization table for character image and with a second quantization table for half-tone image. Image data are discriminated as character image data or half-tone image data according to the quantization data. The quantization data corresponding to the discrimination are subjected to entropy coding. Image data can be divided into first processing blocks of m*m pixel matrices and second processing blocks of M*M pixel matrices where M>m. A second processing block includes one first processing block and adjacent pixels. Image data in first and second processing blocks are subjected to orthogonal transformation to coefficient data. Coefficient data of the second processing block indicates whether the image is a character image, a dot image, or a half-tone image, permitting selection of an appropriate quantization table. Image data of the first processing block is quantized according to the selected quantization table and quantized data are coded. In order to discriminate a dot image, two discrimination windows are provided, with the first consisting of coefficient data around the DC component and the second consisting of other data. Orthogonal transform coefficients of image data are calculated, and maximum absolute values in the windows are obtained. An image is discriminated to be a dot image if the first window maximum is larger than a prescribed threshold value and is larger than the second window maximum.