Improving 3-D LADAR range estimation via spatial filtering

Abstract

A three-dimensional Laser Detection and Ranging (3-D LADAR) system can produce a set of 2-D images with a fast range gate (~2 ns) resulting in a data cube of spatial and range scene data with excellent resolution in both dimensions. Each 2-D range slice image contains the detected photo-electrons at each pixel for a particular range. The photo-electron counts are directly proportional to the return signal intensities incident upon the detector. Range estimation errors of a scene can occur in a 3-D LADAR due to several system factors including the optical spatial impulse response, photon noise, and atmospheric distortion. These factors cause the scene's intensity to spread, or blur, across pixels. The intensity spreading corrupts the correct pixel intensities at each range gate by mixing intensities with neighboring pixels thereby providing false intensity values to the range estimator. Without blur compensation, the range estimates would then be inaccurate to a degree depending on the blur severity. The focus of this paper is to improve 3-D LADAR range estimation by implementing 2-D image restoration filters to "de- blur" each detected 2-D range slice image. Due to simplicity and quickness, this research effort implements two linear image restoration filters (Wiener and inverse filters). Considering the blur due to the optical system impulse response only, implementing the filters on the blurred data shows nearly complete recovery of the correct ranges. The associated root mean square error (RMSE) improves from 0.5 meters before filtering to 0.26 meters after inverse filtering. With typical noise power and moderate atmospheric effects, range estimation improves from a RMSE before Wiener filtering of 0.54 meters to 0.29 meters after filtering with slight degradation to image quality. With typical noise power and light turbulence, range estimation improves from a RMSE before filtering of 0.50 m to 0.28 m after filtering.