An apparatus comprises three-dimensional object detecting means for detecting a three-dimensional object in a taken image, temporarily determining means for calculating, based on parallax information, a lower end height from a road surface to a lower end of the object, for temporarily determining that the object is a ground-contacting object when the lower end height is less than or equal to a threshold, and for temporarily determining that the object is a floating object when the lower end height exceeds the threshold, and finally determining means for finally determining whether the object is a ground-contacting object or a floating object. The finally determining means finally determines that the object is a ground-contacting object at the current computation timing when the object is determined to be a ground-contacting object in at least one of the temporary determination at the current computation timing and the final determination at the previous computation timing.