On The Position Acquiring Algorithm And Experiments For Mobile Targets With Visual Information

Machine vision appeared at the end of the 1960s. With the rapid development of the computer and electronic technique, machine vision has been used to more and more practical applications. Vision information has been used to orientate manipulator for the research from 1970s, applied to their representatives for IC manufacturing, precision electronic product assembly visual systems. After the 1980s, with special image processing hardware emerging, visual information can be used for continuous feedback. The control form based on visual information appeared, and that is known as visual servoing. By 1990s, with the computer's ability to continuously enhance and lower prices, and the rapid development of the image processing hardware and camera technology, visual servoing system has become a major research focus. In recent years, machine vision and visual servoing system in theory and application have been tremendous progress.Ball and plate system is an important research platform for visual servoing system. The ball and plate system acquires the current position of the ball with acquiring and processing images obtained by the camera. Then the system controls the tilt angle of plate with the operation of controller in order to make the ball move to the designated position. Finally, the system completes the task of trajectory control of the ball. The ball and plate system can be simulated robot hand-eye coordination process. In the process the visual information processing, control methods, kinematics and dynamics can be studied. Thus the system can be used as research platform of machine vision and visual servoing. So far, a number of universities at home and abroad have already produced the type of experimental device. The discussion group has built ball and plate system experimental device named BPVS-JLUⅠ. The system has achieved the trajectory control tasks based on visual information. However, with the requirements of high-speed movement of the ball, the original visual information feedback system has been difficult to meet the accuracy requirements. Therefore, the development of a more accurate visual information feedback system has become an urgent goal.From improving accuracy, effectiveness, stability, real-time and adaptability of visual information feedback, this paper design and develop image processing module which is based on digital camera by using a new industrial digital camera. This paper also study on camera parameters automatically adjustment and image processing technology of ball and plate system in the complex environment of light. Title of the dissertation is On the Position Acquiring Algorithm and Experiments for Mobile Targets with Visual Information, and study on mainly image processing. The paper is made up of four parts: 1. The automatic adjustment algorithm of the camera parameters based on mean value of image gray. 2. The dynamic child-image algorithm based on location estimation. 3. The real-time image data acquisition and display 4. The acquiring ball position algorithm under the influence of shadows and glisten.The visual servoing system usually susceptible to outside influence complex light conditions, resulting in decreased performance stability. Digital cameras have more adjustable parameters and scope of the larger adjustable; in common under different light conditions have been satisfied with the level of brightness of the image. For the characteristics of digital cameras, in this paper the automatic adjustment algorithm of the camera parameters based on mean value of image gray has been advanced. The algorithm can automatic adjust the camera parameters under the larger illumination in order to make the images obtained with the appropriate level of image brightness. Visual servoing system similar problems exist. The study has a wide application range.New digital cameras have greater resolution and faster image acquisition speed, leading to volume of processing data in a unit time increasing, which brings the real-time problem of the image processing system. To solve the problem, the dynamic child-image algorithm based on location estimation has been advanced. The algorithm can significantly reduce the time needed for image processing, improve the real-time system, and lay a good foundation for the late image processing.In this paper, image acquisition system in the process of collecting images has spent lots of time. Using a callback function mechanism, the process of image acquisition will be controlled by hardware, and through multi-cache serial transmission of data parallel processing means to achieve the image acquisition, display and processing carried out in parallel. The method for the follow-up of image processing and control saves time and improve the real-time system.Through the image histogram analysis which is collected by ball and plate system, the reasons for the low efficiency of method of threshold extract have been founded. To solve the problem, the dynamic transform enhancement algorithm has been advanced. The algorithm can increase the contrast between the ball pixels and the other pixels. The algorithm also can eliminate disadvantage influence of large background pixels in threshold extract. Against influence of glisten for the ball position accuracy, a multi-projection algorithm has been advanced in this paper. Through this method coordinates of the center of the ball can be extracted. Experiments confirmed that the method has higher accuracy and good adaptability for glisten.The software application system is developed with Visual C++. Functions of the algorithms proved to be reliable and robust. With the specific data and images in actual process of automatic adjustment of the camera parameters, validity of the algorithm has been validated particularly and specifically. The estimated statistical errors which are under different light conditions have been calculated. Experiments proved that the use of the method compared to the entire image and dynamic child-image algorithm improve system performance. Similar to the experimental process of a sampling cycle specific examples verified that the ball position acquiring algorithm is effective.