Binary Weighted Averaging of an Ensemble of Coherently Collected Image Frames

Abstract

Recent interest in the collection of remote laser radar imagery has motivated novel systems that process temporally contiguous frames of collected imagery to produce an average image that reduces laser speckle, increases image SNR, decreases the deleterious effects of atmospheric distortion, and enhances image detail. This research seeks an algorithm based on Bayesian estimation theory to select those frames from an ensemble that increases spatial resolution compared to simple unweighted averaging of all frames. The resulting binary weighted motion-compensated frame average is compared to the unweighted average using simulated and experimental data collected from a fielded laser vision system. Image resolution is significantly enhanced as quantified by the estimation of the atmospheric seeing parameter through which the average image was formed.

Image deblurring and near-real-time atmospheric seeing estimation through the employment of convergence of variance

“…Regardless of the class of sensor and associated noise distribution, a common technique for improving the signal to noise ratio (SNR) of the received image is the process of multiple frame averaging. 2 Many imaging systems currently employed or in development have high enough frame rates to capture and produce multiple frame image ensembles consisting of 10 to perhaps hundreds of images over a short period of time. With proper registration, the summation of these images will tend to average out the effects of noise and thus improve the overall SNR.…”

“…The frequency representation of the PSF in an average sense is the product of the diffraction limited optical transfer function (OTF) and the atmospheric blur. 2 Finally, when using motion compensated frame average (MCFA) images comprised of the summation of numerous short-exposure frames, the atmospheric blur can accurately be described using the shortexposure OTF. 3 A key advantage associated with the employment of the average short-exposure model is that the entire blurring function is reduced to a single unknown parameter (Fried's seeing parameter-r 0 ) which this research will focus on accurately identifying.…”

“…The problem we are attempting to solve was previously addressed by MacDonald. 2,12 This work by MacDonald focused on undoing the distorting effects of atmospheric turbulence in short exposure, partially coherent imaging of remote scenes at long slant ranges. Key contributions of the algorithm originally presented by MacDonald are as follows.…”

Image restoration technique for motion-compensated frame averaged data collected by 3D flash ladar imaging system

“…His work focused on undoing the distorting effects of atmospheric turbulence in short-exposure, partially coherent imaging of remote scenes at long slant ranges. 10,11 By reducing the complexity of the blurring function, he was able to develop an algorithm more suitable for employment in a tactical environment. This research will show that the addition of range diversity in 3-D images allows for simultaneous estimation of the parameterized PSF and spatial / temporal enhancement of the image without the assignment of a prior distribution.…”