An odd- and even-numbered pixel level difference correction process with respect to a linear image sensor is carried out by a successive three pixel weighted moving average calculating unit (71) according to the equation: Sc = (S + U)/2 = {S + (Sp + Sa)/2}/2 = (Sp + 2S + Sa)/4 where S represents a pixel signal to be corrected, Sc a corrected pixel signal, Sp a preceding pixel signal which precedes the pixel signal to be corrected, Sa a following pixel signal which follows the pixel signal to be corrected, and U an average of the preceding and following pixel signals. The odd- and even-numbered pixel level difference correction process can produce corrected image data whose waveform center of gravity remains positionally unshifted unlike a conventional correction process which corrects a pixel signal using only the pixel signal to be corrected and the preceding pixel signal.