Corresponding points are searched based on an error between light beam vectors projected on a projection plane without performing comparison between pixel values at the corresponding points. The necessity for use of cameras having a same camera lens or a same distortion parameter is eliminated, and picked up images of different camera models can be connected to each other. Since original picked up images are pasted directly to an output frame based on errors between light beam vectors without transforming any picked up image once into a pinhole image, deterioration of pixels can be suppressed. Accordingly, picked up images of various cameras which are different in terms of the lens distortion or the camera model from each other can be suitably pasted together.