Patent US 5,301,265

In an initial set up sequence, N pels of an original image are stored in N memory positions of an input buffer. The N pels are converted to M pels, which are stored in M memory positions of an output buffer, where M and N are integers larger than 0. In the case of N<M, an enlargement of the original image is made, and in the case of N>M, a reduction of the original image is made. An address counter pointing to an address of a memory position of the input buffer is initially reset to 0. An address counter pointing to an address of the memory position of the output buffer is also reset to 0. The values -N, M, and an initial sum value S (where S is integer satisfying -N.ltoreq.S<M) are initially stored in three registers, respectively. The initially stored sum value S is repeatedly updated to control the scaling up operations or the scaling down operations. There is a first device which determines or checks whether the sum value is negative, or not. The first device generates a first signal indicating that the sum value is not negative or a second signal indicating that the sum value is negative. A second device responds to the first signal to store or move the pel data in a memory position of the input buffer pointed to by the address counter into a memory position of the output buffer pointed to by the address counter, to update the sum value by adding the value -N to the sum value and to increment the address of memory position of the output buffer; and which responds to the second signal to update the sum value by adding the value M to the sum value, and to increment the address of the memory position of the input buffer. A control device is provided to repeatedly control the above operations until the M pels of the scaled image are formed.