Pose robust human detection and tracking method using the stereo image

摘要

human detection method using stereo images of the present invention is generated using sigakcha information from stereo images , a two-dimensional space - that is converted from a two-dimensional histogram sigakcha space - depth detecting a histogram of the candidate of the person through operation seungjeok directional two-dimensional elliptic filter of the four directions, with respect to the detected face candidate person detection and head - by performing the validation through the shoulder shape matching , if the angular pose of a person in the front stereo image , not to provide people with high detection rate .