Research On Data Matching Algorithm Based On The Feature Extraction Of Chamfer Distance

In order to reflect the integrity of the entity model of point cloud data, the point cloud matching technology which based on the reverse engineering has been widely used. In this paper, the point cloud matching technology and its algorithms are deeply researched. On this basis, data matching algorithm based on the feature extraction of chamfer distance is pointed out to improve the matching precision.The major work which completed in this paper is as follows:1. An improved ICP algorithm based on curvature is inspected and verified. Firstly, according to the geometric information of measured point and each point in the neighborhood, surface fitting was discussed. Then the curvature of the point on the curved surface was estimated. Based on the similarity of curvature, the characteristic points were searched for matching computation. The results of experiment showed that the matching effect is defective and the area of overlapping shadow was found.2. On the base of point cloud preliminary matching with PCA, a data matching algorithm based on feature extraction of chamfer distance is proposed to further improve the point cloud matching precision. Above all, the algorithm used geometric information of point cloud data to calculate the eigenvectors of the covariance matrix of measuring point, and realized the preliminary match of point cloud with space coordinate transformation. Firstly, using chamfer distance transforms method to search the feature point and determine its corresponding point set. And then, the least square method is used to calculate the error function iteratively. Finally, SVD is used to solve coordinate transformation. The algorithm realizes high precision matching of point cloud data.3. The real-time operation of algorithm is realized with Geomagic and programming software. According to the results of the experiments, the algorithms have mentioned in this paper are analyzed and compared. The results showed that point data matching algorithm based on the feature extraction of chamfer distance has a good matching effect.