Pedestrian extracting apparatus

摘要

A pedestrian extracting apparatus of the present invention extracts a whole-body region 8 and a prospective head region 9 based on a pattern matching between an input image 3 picked up by an infrared camera 2 and whole-body/head model images 6 and 7 stored in a model-image storage unit 14. If the whole-body region 8 has not been extracted, the prospective head region 9 which is at the closest position to a prospective head region 9c extracted from a preceding input image 3' is extracted from the current input image 3, and an image matching with the head model image 7 of the extracted prospective head region 9 is determined to be a head.