Method for object pose recognition of robot by using tof camera

The present invention relates to a method for recognizing an object posture of a robot using a time of flight (TOF) camera, and to acquire a distance information image of an object to be recognized using a TOF camera, After detecting the edge image, the distance information image and the edge image are ANDed to obtain 3D point cloud data consisting of effective feature points of the object to be recognized, and stored in the database. The posture of the object to be recognized is recognized by matching with three-dimensional point group data of the entire object to be recognized. According to the present invention, even in the case of non-visible objects such as glass or transparent acrylic plates, high-quality three-dimensional point group data can be obtained, and by using three-dimensional point group data (3D Point Cloud) composed of effective feature points (Constrained Features) As a result, the work time can be drastically reduced when the iterative closest point algorithm is executed. Robot, Object Pose Recognition, TOF Camera, 3D Point Cloud, Feature Point