In an image acquisition device, an optical path difference generating member can form an optical path length difference of a second light image without splitting light in a second optical path. This can suppress the quantity of light required for the second optical path to obtain information of the focal position, whereby a quantity of light can be secured for a first imaging device to capture an image. The image acquisition device synchronizes the movement of a predetermined part of a sample within a field of an objective lens with rolling readout such that each pixel column of a second imaging device is exposed to a light image of the predetermined part in the sample.