Research On Key Technologies For 3D Reconstruction Of Fruit Tree's Stems

With the development of computer technology and the practical demand of farm production,3D reconstruction of fruit trees has become a hot topic.In this dissertation some key technologies of 3D reconstruction have been studied.A series of mathematical models have been offered mainly aiming at image points,line feature matching algorithm,camera demarcating,model validating method of 3D reconstruction.Meticulous discussion on the localization and deficiency of classical method has been carried out.The improved technology has been presented and applied to the field of 3D reconstruction on the stem of fruit trees.Experimental results have been obtained through computer operation.The research contents and achievements of this paper are as follows:Firstly,an improved method of Tsai camera demarcating based on laser orientation has been offered and applied to the field of 3D reconstruction on the stem of fruit trees.It improved on error in classic methods owe to the imprecise scalar and carried out easily.Using this method,accurate results were gotten.Secondly,aiming at the deficiency of the feature matching algorithm in 3D reconstruction based on image,an improved matching algorithm based on image feature correspondence has been offered and applied to this field of the stem of fruit trees.Using the maximum-weight bipartite graphs technology, we get the maximum correspondence across images and the veracity for 3D coordinate of features in imagines and 3D reconstruction would be ensured.Through this method,image feature matching and 3D reconstruction could be operated in parallel and the limitation of traditional method of polar line restrict which measures features from 2D neglecting 3D information.In a word,the method that could combine calculating affinity measure and building a maximum-weight bipartite graph would make image feature matching more convenience and more reliable.The matching accuracy and speed were improved.As the complexity for 3D reconstruction of tree,it was hard for judging the effect for reconstruction. In this passage a method and algorithm for reconstructing the three-dimensional surface of tree stems using laser scanner was presented.Through this way,the model of tree could be validated and the problem of judging the effect for reconstruction would be solved.Finally,in order to check up these key technologies presented by the passage,some experiments using these technology were carried out and the result gotten by experiments were compared with data through other methods.Experimental results on synthetic and real data demonstrate the effectiveness of these key methods.