Research On3D Point Cloud Map-building Based On TOF Camera

Application that3d camera implementation of the principle of3d point cloud mapbuild based on the Time Of Flight method (TOF) is one of the hot spots of currentresearch abroad, The study can be established for the robot map, navigation, positioning,three-dimensional environment target recognition. According to the situation thatdomestic research on this technology is less, this paper does research on this hot spot andachieve the aim of environment of3d reconstruction through the technology.According to the characteristics of the application of the technology, At first, thispaper calibrated gray image by Zhang ZhengYou method to improve the accuracy of3dreconstruction and according to the different factor affecting the accuracy of distancedata, proposed the error compensation method.After the compensation this paper joined point cloud data together at meantimestudied merging algorithm based on feature points matching and algorithm based oniterative closest point(ICP) algorithm. The main course of the first algorithm firstlymatched the feature points extracted between two frame gray images, applied a variety ofmethods for filtering out the pairs of feature points which were matched by mistake, thengot the relative pose transformation matrix by the pairs of feature points correctlymatching to join two point cloud together. Improved the efficiency of the ICPalgorithm by the application of kdtree search and down sampling.Through the analysis and experimental comparison of the characteristics ofexperimental between two kinds of merging algorithm, this paper used the mergingalgorithm based on the ICP algorithm that toke results of the algorithm based on thefeature points matching algorithm as the initial value. Through this method joined themultiple frame data together, achieved the purpose of the3d point cloud map building.