Shaded picture signal processing system and method
AssigneeFuji Photo Film Co., Ltd.
InventorSumio Mori
Life of this patent
Estimating the term…
Abstract
A shaded picture signal processing system and method which produces recorded pictures of high quality with a small number of gradation levels with the resolution of the picture maintained nearly unchanged. False contours appearing as quantization distortion are removed by dispersing noise of equal magnitude to the false contours throughout the whole picture or a specified portion of the picture so that the false contours cannot be detected by the human eye.
Claims5
A shaded picture signal processing method for obtaining an output picture signal by quantizing an input picture signal having gradations of p bits per picture element into gradation levels of q bits per picture element with said input picture signal formed in a block of m.times.n picture elements, comprising the steps of:
- calculating a quantized value x.sub.qij by quantizing in q bits per picture element x.sub.ij in a spatial address (i, j) in said block;
- calculating a higher quantization value x.sub.uij by adding predetermined quantization step data to said quantized value x.sub.qij ;
- calculating a quantization error e.sub.ij which is a difference between said picture element x.sub.ij and said quantized value x.sub.qij ;
- and providing a threshold pattern in m.times.n matrix form in correspondence to said one block;
- selecting one of said quantized value x.sub.qij and said higher quantization value x.sub.uij in response to a comparison of said quantization error e.sub.ij with a threshold c.sub.ij at a spatial address (i, j) in said threshold pattern to output an output picture signal y.sub.ij of q bits per picture element corresponding to said spatial address thereby to reproduce halftones.
A shaded picture signal processing method in which an input picture signal is formed in a block m.times.n picture elements, comprising the steps of:
- calculating an average value x from picture elements x.sub.ij of said block;
- dividing said block into a region A in which picture elements x.sub.ij .gtoreq.average value x and a region B in which picture elements x.sub.ij <average value x;
- calculating average values x.sub.1 and x.sub.0 of said picture elements x.sub.ij in said regions A and B;
- calculating quantized values x.sub.1q and x.sub.0q by quantizing in q bits said average values x.sub.1 and x.sub.0 ;
- calculating higher quantization values x.sub.1u and x.sub.0u obtained by adding said quantized values x.sub.1q and x.sub.0q to predetermined quantization step data;
- calculating quantization errors e.sub.1 and e.sub.0 which are differences between said average value x.sub.1 and said quantized value x.sub.1q and between said average value x.sub.0 and said quantized value x.sub.0q ;
- providing a threshold pattern in m.times.n matrix form in correspondence to said block;
- and obtaining an output picture signal y.sub.ij in which each picture element comprises q bits corresponding to a spatial address (i, j) in said threshold pattern by selecting one of said quantized value and higher quantization value by comparing said quantization error with a threshold c.sub.ij in said spatial address, wherein, when said spatial address (i, j) is in said region A said quantization error e.sub.1, quantized value x.sub.1q and higher quantization value x.sub.1u are employed and when said spatial address is in said region B, said quantization error e.sub.0, quantized value x.sub.0 q and quantization higher value x.sub.0u are employed;
- and outputting one of said quantized value x.sub.1q and quantization higher value x.sub.1u, and said quantized value x.sub.0q and quantization higher value x.sub.0u to reproduce halftones.
Dependencies are inferred from the claim text.