When a mobile is detected using an imaging device installed in the mobile, the image of a partial area is enlarged/reduced depending on variation in distance to the detection object mobile and then it is compared under a fixed scale thus causing increase in computation cost. In order to eliminate the need for an enlargement/reduction processing or a deformation correction processing every time when collation is performed, an input image is converted into a virtual plane image having a size or a shape on the image of a detection object mobile which does not vary depending on the distance between the mobiles. Using a pair of virtual plane images obtained at two different times, points are made to correspond and the mobile is detected based on the gap of corresponding points.