Optical Correlation Based Pose Estimation Using Bipolar Amplitude Spatial Light Modulators
Final rept. 1 Apr 2007-31 Jul 2008
ALABAMA UNIV IN HUNTSVILLE
Pagination or Media Count:
Pose estimation using a bank of filters and a pose search algorithm approach was performed in both the digital and optical domains sub-degree standard deviations in accuracy. The digital routine required less image processing than the optical method. This is because of this inability of the input SLM to achieve the resolution of the original digital image. The digital method is also capable of implementing complex composite filter designs. These filters generally performed poorly when directly optically implemented. This was not an unexpected result ar was due to the limited dynamic range of the filter SLM and its amplitude only real valued modulation. Using Matlab as an interface to the correlator for the pose estimation routine and for the development of filter conversion algorithms and image processing of the 3D software generated training images allowed for the programming and generation of the filter bank and image loading sequence for bank of filters with search algorithm proof of concept. This allowed for programmable changes to filter designs or image processing to be implemented within a single software platform. This general approach to pose estimation is therefore more streamlined and less costly in time and hardware from development to simulation than other existing approaches.