Registration of 3 D-LiDAR Data With Visual Imagery Using Shape Matching

Multi-modal image registration plays a central role for many applications. In this paper we establish a new method for multi-perspective registration of 3D dimensional Light Detection and Ranging (LiDAR) data and visual imagery. The proposed method consists of three steps. In the first step, common features are identified and extracted from both modalities. Two dimensional objects are extracted from both modalities. We introduce a new method for LiDAR segmentation and object boundary extraction. Similarly we extract objects from visual images using k-means segmentation algorithm. The second step is features matching and and determination of corresponding points. In order to match the objects extracted from both modalities, a position, scale and rotation invariant shape signature is used to represent each object. A bipartite graph consisting of two sets, LiDAR objects and visual objects is constructed. A histogram similarity metric is used to assign a matching cost between every pair of objects signatures. This graph is used as input to the Hungarian algorithm to find the best matching. After determining the best matching objects, a set of matching points is selected using objects with lowest matching cost. In the third step, these points are used to compute the mapping between the two modalities. Experiments are conducted on synthetic data and on real world data. we also introduced a new metric for computing the registration error using dynamic time warping.

[1]  A. Habib,et al.  Photogrammetric and Lidar Data Registration Using Linear Features , 2005 .

[2]  Kiyun Yu,et al.  Registration of aerial imagery and aerial LiDAR data using centroids of plane roof surfaces as control information , 2006 .

[3]  Avideh Zakhor,et al.  Automatic registration of aerial imagery with untextured 3D LiDAR models , 2008, 2008 IEEE Conference on Computer Vision and Pattern Recognition.

[4]  Kenneth Steiglitz,et al.  Combinatorial Optimization: Algorithms and Complexity , 1981 .

[5]  David G. Lowe,et al.  Object recognition from local scale-invariant features , 1999, Proceedings of the Seventh IEEE International Conference on Computer Vision.

[6]  Venu Madhav Govindu,et al.  Alignment Using Distributions of Local Geometric Properties , 1999, IEEE Trans. Pattern Anal. Mach. Intell..

[7]  Jan Flusser,et al.  Image registration methods: a survey , 2003, Image Vis. Comput..

[8]  Bernard Chazelle,et al.  Shape distributions , 2002, TOGS.

[9]  Siamak Khorram,et al.  A feature-based image registration algorithm using improved chain-code representation combined with invariant moments , 1999, IEEE Trans. Geosci. Remote. Sens..

[10]  Matthijs C. Dorst.  Distinctive Image Features from Scale-Invariant Keypoints , 2011 .

[11]  John W. Fisher,et al.  Automatic registration of LIDAR and optical images of urban scenes , 2009, CVPR.

[12]  David A. Clausi,et al.  Automatic registration of SAR and visible band remote sensing images , 2002, IEEE International Geoscience and Remote Sensing Symposium.

[13]  Billy E. Gillett,et al.  Introduction to Operations Research: A Computer-Oriented Algorithmic Approach , 1976 .

[14]  Ayman Habib,et al.  CO-REGISTRATION OF PHOTOGRAMMETRIC AND LIDAR DATA: METHODOLOGY AND CASE STUDY , 2004 .

[15]  Tania Stathaki,et al.  Shape Signature Matching for Object Identification Invariant to Image Transformations and Occlusion , 2007, CAIP.