The Research On Binocular Visual System Of The Snake-like Probe

In the case of accidents in mines or in earthquake, as the space is small, rescue workers is difficult to locate and rescue the wounded. Therefore, a snake-like robot is introduced for the detection and rescue work. Based on adequate research of binocular visual modeling technology, binocular vision control method of the snake-like robot has been researched and a snake-like robotic binocular vision system has been built for this purpose.Modular snake-like robotic system has been studied, and the system design has been improved. By study of snake-like robot control method based on binocular stereo visual, a platform for experimental research on the theories of binocular stereo visual has been built. With the theories of binocular vision, image algorithms processing, binocular vision matching, camera calibration,3D reconstruction problem of the snake-like robotic binocular vision system has been discussed in detail.The basic image extraction and image processing algorithms is developed. The maximum value greater than the threshold in the scale-space was chosen as the SIFT feature points by using the SIFT algorithm. Stereo matching on the feature points based on characteristics of the scale-invariant was carried out with related matching constraints.Calibration method for stereoscopic cameras is studied by using binocular vision and camera distortion models. Camera calibration experiment with Zhang algorithm is used to obtain the internal and external camera parameters.3D coordinate system is constructed by experiment for camera calibration of experiment platform. This3D coordinate system can be used to calculate the3D coordinates that those characteristic points in matches corresponded and built the3D simplified map of the surroundings.Selecting Visual Studio2010environment and Open CV tools, software including image capture, SIFT detection, camera calibration,3D reconstruction system is developed.