Research Of Indoor Property Protection Based On RGB-D Image

With the development of security monitoring technology,more and more security technology products are widely used in daily life,supporting our lives and work convenience and protection.Existing indoor property protection technology is mainly about the foreground extraction from color video images to detect the removed objects,and the object tracking or facial recognition to find a suspect.Along with the emergence and development of 3D technology in recent years,a variety of devices representing by Kinect of Microsoft Corporation which can obtain the depth image appeared,the research on RGB-D image increased,providing a new opportunity for indoor property monitoring.Based on the research of traditional indoor property protection technology and extraction of RGB-D image by Kinect,removed object detection is carried out to obtain the information of stolen object,the locking of suspect is completed by matching carried object to removed object,and the human identification is achieved through bone tracking.The main contents are as follows:A method to retrieve information of removed object detection with RGB-D image was proposed.In the color image,the foreground was extracted by modeling the background with GMM.Meanwhile,depth change information was obtained in depth image.The area of removed objects was finally located by update different depth change area with different parameters.The proposed method can overcome the shortage of traditional methods and can effectively divide the area of removed objects,people and people with baggage.Through experiments in the indoor environment,the method proposed in this paper can detect the removed objects effectively with good adaptability.In order to further determine the suspects of the removal objects,a detection method based on the human joint nodes and feature matching is proposed.Based on the analysis of the status of human with carried objects,the possible area of the carried objects can be obtained by positioning the key bone nodes.Through the match with the removed objects by SIFT algorithm to lock the suspect and achieve indoor property protection.Finally,the experiment is verified and analyzed.The results show that the method can effectively extract the information of the carried objects and can be used to identify the suspect.In order to lock suspects between legal users and illegal users,an identification method based on skeletal information is proposed.Through the human skeletal data obtained by Kinect,the distance information of the human skeleton joint point is calculated.By matching with the distance information of legal users in database,human identity is confirmed.Test results show that the method is accurate and can distinguish the user 's identity effectively in the indoor environment.