According to the embodiment described above, in RC1-RC3 in the most upstream leading-end position detecting process RB1, the pixel positions, which are apart upstream from one another for the predetermined quantity of number, in the range along the main-scanning direction MD between the central pixel position and the foremost pixel position are appointed as the objective position TMDP until the edge is no longer detected in the objective position TMDP. In RC5 and RC6, the pixel positions in the range between the objective position TMDP, in which no edge was detected, and the objective position TMDP, in which the edge was detected, are appointed one by one downstream along the main-scanning direction MD as the search position DMDP. In RC7, the search position DMDP, in which the edge was detected, and the edge position EP of the search position DMDP in terms of the conveying direction FD are saved as the most upstream leading-end position UHP. The most downstream leading-end position DHP is saved in the similar manner as the most downstream leading-end position detecting process RE1. Thus, after the objective positions TMDP are appointed intermittently at the interval of the distance equivalent to the predetermined quantity of pixels, the search positions DMDP are appointed one by one sequentially. Therefore, the most upstream leading-end position UHP and the downstream position DHP may be detected promptly and accurately. In this regard, even when the original sheet GS skews, and the image reader 24 starts reading the leading end area of the original sheet GS from a midst position in the conveying direction FD, the width of the original sheet GS, i.e., the size of the original sheet GS in the main-scanning direction MD, may be accurately determined.