Real Time 3D Reconstruction And Pose Estimation Of Wheel Based On Dual Semicircular Planar Target And Rotating Laser

In this paper,an approach of performing calibration and 3D reconstruction simultaneously based on dual semicircular planar target and rotating laser is proposed.A turntable or a linear positioning system is not necessary in this method to obtain the complete 3D reconstruction data compared to conventional methods.The approach is used to scan the wheels to get the cloud points,these cloud points then be applied to get the four wheel alignment parameters.Since there is no need adding a precise planar target to the wheel hub,it enhances the convenience of doing fourwheel alignment.Firstly,we do some research on the theory of cameras calibration,we put much effort to study the method put forward by Zhang.The capability of using a semicircular planar target to calibrate the cameras is discussed.Then the methods of extracting the medium axis of laser light stripe are explored.When the light stripe is projected to surface with strong diffuse reflection,many conventional methods of extracting skeleton will fail to get the exact medium axis.After researching the theory of fuzzy distance transform,we proposed a method that combines the theory of fuzzy distance transform and the methods of skeleton pruning to get the precise skeleton of those light stripe and better skeleton result is got.Then we study the theory of calibrating the plane parameters of structured-light measurement device.We focus on the method on calibrating the device with only one laser projector and one camera.After doing research of how to obtain the complete 3D reconstruction of objects,we proposed a novel design using a dual semicircular planar target and a rotating laser projector to perform calibration and 3D reconstruction simultaneously.Since the laser is rotated 360 degrees,the whole surface of object can be scanned,so a turntable or a linear positioning system is not needed to obtain the complete cloud points.We scanned a plastic model to test the approach and good 3D reconstruction result is got.Then we put efforts to study the theory of four wheel-alignment and maintain the details of calculating the four angles in four wheel-alignment.The 3D scan approach proposed above is used to scan the wheels,after getting the cloud points a fusion method is taken to make all points lie in the same coordinate system,then four angles in four wheel-alignment relative to the new coordinate system can be worked out.In the last part,experiments are conducted to validate the flexibility and the accuracy of the system.Good 3D reconstruction result is obtained and precise four wheel-alignment parameters are got.