Improved registration for 3D image creation using multiple texel images and incorporating low-cost GPS/INS measurements

摘要

The creation of 3D imagery is an important topic in remote sensing. Several methods have been developed to create 3D images from fused ladar and digital images, known as texel images. These methods have the advantage of using both the 3D ladar information and the 2D digital imagery directly, since texel images are fused during data acquisition. A weakness of these methods is that they are dependent on correlating feature points in the digital images. This can be dicult when image perspectives are signicantly dierent, leading to low correlation values between matching feature points. This paper presents a method to improve the quality of 3D images created using existing approaches that register multiple texel images. The proposed method incorporates relatively low accuracy measurements of the position and attitude of the texel camera from a low-cost GPS/INS into the registration process. This information can improve the accuracy and robustness of the registered texel images over methods based on point-cloud merging or image registration alone. In addition, the dependence on feature point correlation is eliminated. Examples illustrate the value of this method for signicant image perspective dierences.