OPTICAL THREE-DIMENSIONAL SURFACE RECONSTRUCTION AND EVALUATION USING A MULTIMODALITY IMAGING INSTRUMENTATION

摘要

This paper proposes an optical surface reconstruction method from multiview projectional data acquired by a multimodality imaging instrumentation. The technique is adapted for in vivo small animal imaging, specifically imaging of nude mouse in the case where the influence of CT radiation doses should be eliminated. Any potential point within the field-of-view (FOV) of a nude mouse is evaluated by a proposed photo-consistency measure utilizing sensor image information. As the superposition of adjacent projections yields depth information for any point within the FOV, the three dimensional (3D) surface of the imaged object is estimated by a graph-cuts based method through global energy minimization. The reconstructed surface is evaluated by comparing the the reconstructed surface and the CT volume of the nude mouse. The proposed surface reconstruction method demonstrates the feasibility of surface reconstruction from multiview projection, and there are still great possibilities to improve this method.