An image processing apparatus includes a recorder. A recorder records repeatedly-fetched object scene images as a moving image. An extractor extracts one portion of the object scene images recorded by the recorder, in parallel with the recording process of the recorder. An evacuator alternatively evacuates a predetermined number, which is an upper limit, of the object scene images extracted by the extractor. An allocator allocates the object scene images evacuated by the evacuator at a time point at which the recording process of the recorder is ended, to the moving image recorded by the recorder. A controller starts up the evacuator at a frequency that is decreased according to an increase in an amount of moving images recorded by the recorder.