Development Of Virtual Swimming Training Simulating Device

As a sport of great antiquity,swimming plays an important role in public sports and athletic sports.It's an important part of our sports power strategy.At present,the public swimming fitness campaign is facing many problems,such as water pollution,potential safety hazard of outdoor public waters and insufficient capacity,worrying health condition of indoor swimming pools.Although swimming simulators can be used for training in competitive swimming,there are still problems such as attitude evaluation difficulty and lack of reality from the perspective of scientific training.Based on the above situation,the virtual reality technology is introduced into the swimming project,and a virtual swimming training simulating device is developed.The main work is summarized as follows:1.Overall scheme.Aiming at the technical requirements of virtual swimming training simulating device,the overall design scheme of the device is determined.The design and installation and strength check of the supporting structure are carried out from the perspective of the comfort and safety of the virtual swimming.The design and selection of hardware,including the design of locating ball set,the choice of camera and the selection of VR HMD,are carried out to satisfy the requirements of.pose detection scheme and virtual environment display.The software function designs of pose detection program and virtual scene program are carried out from the requirements of realizing natural human-computer interaction.2.Pose detection method.An image processing method,solving PnP problem is determined to detect pose.Firstly,the calibration method of Zhang Zhengyou for obtaining the parameters of camera are studied.Secondly,according to the characteristic that effective information accounts little in the motion picture,the method of target identification is studied.The principles of LBP,Haar and HOG image features are analyzed,the method of multi-scale sliding detection using cascade Adaboost classifier is proposed to extract the region of locating ball set.Finally,the color recognition method for determining the coordinate correspondence and the morphological filtering and blob detection method for calculating the image coordinates of feature points are studied.3.Head-pose detection software.Camera calibration are finished using Zhang Zhengyou calibration method and the pose and position of the current camera in fixed ground coordinate system is obtained using the reprojection method.The classifier training are accomplished based on OpenCV.The performance comparison test of classifiers using different image features is finished and the cascade Adaboost classifier using HOG features is finally chosen to perform the target recognition.Based on the Visual Studio 2010 development environment,the head-pose detection algorithm is implemented and the pose detection program is developed.4.Virtual environment software.Based on Unity3 D software,the virtual scene program is developed and finally deployed on the Android phone in the VR HMD,setting up the initial menu scene and the virtual swimming scene,using c# script to achieve the data transmission,coordinate transformation and visual field updating.5.Experiment and test.The real-time test and accuracy experiments are carried out and the experimental data is analyzed.The instantaneity of updating the visual field and the accuracy of head-pose detection are verified in the virtual swimming device.The virtual swimming training simulating device can simulate the real scene swimming environment,and provide the whole real swimming experience based on virtual reality technology.It's not only safe,but also real-time,interactive and comfortable.While introducing a new sports mode for public swimming fitness and promote the scientific training level of athletic swimming program,it's also of great significance in improving the social recognition for virtual reality applications.