Vision Service Support System Of Intelligent Space In The Hospital Evironment

With the help of smart space, using service robots in hospital ward patrol can greatly reduce the workload and mental pressure of nurses. It can also be able to notice the problems of the hospitalization and improve patient satisfaction. A vision service support system of intelligent space in the hospital evironment is proposed in this thesis. This system uses a variety of computer vision methods to achieve video surveillance and help service robots patrol in hospital. The main contribution of this thesis are shown as follows:Firstly, a global vision system based on cameras set on the ceiling is proposed. After comparison of the commonly used methods for motion detection, a new kind of motion detection method based on Motion History Image (MHI) is proposed. This method is able to set background image when there are moving objects in the vision and is able to quantify the quality of the background. An object localization algorithm using camera intrinsic parameter and extrinsic parameter is proposed. A large number of experiments show that the error of the algorithm is appropriate.Artificial landmark is an important auxiliary method for robot localization and navigation. A kind of new artificial landmark based on QR Code technology is proposed in the paper. This new landmark can not only store rich environment information, but also provide effective pose information to the robot. A fast algorithm based on shape and color of the landmark is presented in order that the landmark can be recognized by the robot in far distance. Then the localization algorithm using vanishing line principle is designed to locate the artificial landmark quickly and accurately in the camera coordinates of the robot. Finally, a control scheme including searching in far distance, recognition, localization and reading landmarks is proposed, which lays a solid foundation for robot autonomous localization and navigation. The experiments demonstrate that the new landmark has far recognition distance, high recognition speed, high positioning precision, high decoding efficiency and stability even in complex indoor environment.A computer vision system on robot based on TI TMS320DM642 is proposed. This system can acquisition, compress, process and transfer real-time digital image data. Server side and client side program is done. Patrol procedure and image-processing algorithm for each key target was proposed. Robot can accomplish patrol missions rapidly and accurately based on this system.