Advanced algorithms for identifying targets from a three-dimensional reconstruction of sparse 3D ladar data

Abstract

There is a considerable interest in the development of new optical imaging systems that are able to give threedimensional images. Potential applications range across medical imaging, surveillance and robotic vision. Identifying targets or objects concealed by foliage or camouflage is a critical requirement for operations in public safety, law enforcement and defense. The most promising techniques for these tasks are 3D laser imaging techniques. Their principles are to use movable light sources and detectors to collect information on laser scattering and to reconstruct the 3D objects of interest. 3D reconstruction algorithm is a major component in these optical systems for identification of camouflaged objects. But 3D reconstruction must take into account sparse collected data i.e. concealed objects and reconstruction algorithms must solve a complex multi-parameter inverse problem. Therefore the inverse problem of recovering the surface three-dimensional shape function from intensity data is more challenging. The objective of our paper is to present a new algorithmic approach for the generation of 3D surface data from 3D point clouds corresponding to reconstruction algorithm. This algorithmic approach is based on research of automatic minimization of an energy function associated with a sparse structure of 3D points. The role of this type of algorithmic data-driving process is to complete the incomplete 3D image at satisfactory levels for reliable identification of concealed objects.