To obtain an exposure time decision device where the possibility of saturation of each color component is precluded while keeping color balance of reproduced images.

Reference exposure times Rs, Gs, Bs are obtained, where R, G, B output data reach a prescribed level in a non-filtering state. Rough reading is made for a prescribed check exposure time (t) (step 202), and a histogram denoting distribution of a signal level of the R, G, B image data is generated (step 203). Valid levels DR, DG, DB smaller by a prescribed amount than a maximum value in each histogram are obtained (step 204). Valid R, G, B levels DR, DG, DB are normalized by using a reference exposure time Gs (steps 205). An optimum exposure time TR is determined from a maximum level among normalized valid levels DRn, DGn, DBn according to an equation of TR =(prescribed value/DR)×t by using the check exposure time (t) (step 207). The optimum exposure times TG, TB of G and B are by using the TR and ratio of Gs/Rs and Bs/Rs (step 208).