A vehicle surroundings monitoring apparatus recognizes a pedestrian from existence of a leg, in the case in which a width in real space of a binarized object is suitable as a pedestrian's leg, the binarized object has longwise shape, the lower end position of the binarized object exists in the lower part (lower half) of a gray scale object region, and a serial luminance transition exists in the binarized object. The vehicle surroundings monitoring apparatus also recognizes a pedestrian from existence of an umbrella held up by the pedestrian, in the case in which an umbrella region with a low temperature exists above the binarized object, the width of the umbrella region is suitable as a pedestrian's umbrella, and, the spatial correlation between the centroid position of the umbrella region and the centroid position of the binarized object is suitable as a spatial relationship between a pedestrian and his/her umbrella.