Research Of Outdoor Street Map Building Maethod

With the three-dimensional equipment constantly improving and map building technology continuously expanding, many big Internet companies’ street view begin stepping into our daily lives, and Providing for user 360-degree panoramic image of cities, streets or other place, and realizes the " Human view" browsing experience. It also has brought great convenience in our daily life, especially in the aspects such as tourism, urban planning, image measurement. How to accurately for robot localization and built figure is an important content in the street view service. This article aims at reconstructing real 3D scene information basing on street images collected by Kinect and IN-RT robot. And we focus on the following aspects.First, we put forward the calibration algorithm between Kinect and the IN- RT robot. Through the robot’s two translational and rotational, we figure out the rotation matrix and translation vector, the rotation matrix and translation vector is from Kinect coordinate to robot coordinate. When calculating the robot’s motion control quantity, we use odometer; When calculating the Kinect’s motion control quantity, we uses KLT algorithm to tracking features and uses RANSAC algorithm to eliminate error points, then use quaternion algorithm to calculate motion control quantity of Kinect. Finally figure out the transformation parameters.Secondly, I was studied using two different methods to solve the problem of robot localization and built figure, and put forward the method of using GPS information to correct locating results. At this part, the paper first puts forward the method of using odometer positioning information and through the coordinate transformation to convert the point cloud data to the global coordinates to solve the problem of positioning and build figure. Then the paper gives the method of using particle filter algorithm solving the problem of location and built figure at the same time.Again, I was studied Point cloud data splicing algorithm and curved surface fitting algorithm. the paper gives using the least squares method and the ICP algorithm solving the problem of point cloud data precise stitching.Finally, we design experiments to prove the accuracy and feasibility of localization and mapping method. And the experimental results are analyzed.