To prevent effective sensitivity from being decreased by arranging a 1st light receiving section and a 2nd light receiving section in parallel while deviating them by a specific length in the vertical direction and delaying an output of a 2nd pixel array with respect to a 1st pixel array by a specific time so as to attain high resolution.

A 1st light receiving section 4 and a 2nd light receiving section 5 in a linear sensor 1 are arranged while being deviated by a size (x) that is a sum of an integer multiple (a) of a size (m) of the 1st light receiving section 5 in the vertical direction and a half of the size (m). Furthermore, an analog memory section 8 delays an output of a 2nd pixel array 3 with respect to an output of the 1st pixel array 2 by a time equivalent to the integer multiple (a) in the deviation of the 2nd light receiving section 5 with respect to the 1st light receiving section 4 in the vertical direction. Thus, a high resolution is obtained the same as the case with reading an original by halving a size of a conventional light receiving section in the vertical direction apparently. In this high resolution processing method, since signal outputs of the 2nd pixel array 3 and the 1st pixel array 2 are synthesized without changing the size (m) of the 1st light receiving section 4 and the 2nd light receiving section 5 in the vertical direction from the conventional size, the deterioration in the effective sensitivity of the linear sensor 1 is prevented.