METHOD AND APPARATUS FOR CONVERTING A LINE DENSITY OF A BI-LEVEL IMAGE SIGNAL

A correspondence relation between a structure of a bi-level pattern A and a structure of a bi-level pattern B is statistically learned by use of a learning high-density image. The bi-level pattern A forms a partial region of a low-density character and line image except a screened dot image. The bi-level pattern B forms a corresponding partial region of a desired high-density image. A judgement is made as to whether the low-density image is equal to or different from a screened dot image. Reference pixels are selected from the partial region of the low-density image. The low-density bi-level pattern A is converted into the high-density bi-level pattern B on the basis of a result of the learning by use of the selected reference pixels when the low-density image is different from a screened dot image. A pixel of the low-density image is repeated to convert the low-density image into the high-density image when the low-density image is equal to a screend dot image.