Digital holography and image sharpening have been used increasingly in recent years for wavefront sensing and imaging. Compared to conventional imaging and wavefront sensing techniques, digital holography and image sharpening require significantly fewer and simpler optical components to retrieve the complex field (i.e., both the amplitude and phase) and produce a focused image from an estimate of the phase aberrations present in the imaging system. A drawback for digital holography in real-time applications, such as wavefront sensing for high energy laser systems and high-speed imaging for target tracking systems, is the fact that digital holography and image sharpening are computationally intensive, requiring iterative virtual wavefront propagation to optimize sharpness criteria. Recently, it was shown that minimum variance wavefront prediction can be integrated with digital holography and image sharpening to reduce significantly the large number of costly sharpening iterations required to achieve near-optimal wavefront correction. This paper demonstrates further gains in computational efficiency with a new subspace sharpening method in conjunction with predictive dynamic digital holography for real-time applications. The method sharpens local regions of interest in an image plane by parallel independent wavefront correction on reduced-dimension subspaces of the complex field in a pupil plane. Results in this paper from wave-optics simulations show that the new subspace method produces results comparable to that from conventional global and local sharpening, and that subspace wavefront estimation and sharpening coupled with wavefront prediction achieves order-of-magnitude increases in processing speed.
Image sharpening is a state-of-the-art phase-retrieval estimation algorithm used in coherent imaging that has demonstrated strong performance under various imaging conditions. However, computational overhead makes these iterative algorithms difficult to enable in real-time and dynamic applications such as high-energy laser and target-tracking systems. Recently, time-invariant, minimum-variance prediction filters have been shown to improve the convergence speed of image-sharpening algorithms during phase retrieval. The research presented here proposes a neural network with on-line adaptive learning to predict a priori wavefront errors for temporally correlated dynamic coherently imaged measurements. The method bootstraps the a priori prediction computed from past samples and updates with the phase-retrieval image-sharpening estimate at the current time step. Testing performed using wave-optics simulations demonstrates that this procedure improves the estimation of dynamic phase retrieval compared to conventional image sharpening, without being bound to a time-invariant constraint. Results also show the capacity for this approach to operate on a cycle that periodically circumvents the process of iterative conjugate-gradient phase retrieval in real-time, achieving order-of-magnitude gains in computational performance compared to current image-sharpening methods.
Digital holography holds several advantages over conventional imaging and wavefront sensing, chief among these being significantly fewer and simpler optical components and the retrieval of complex field. Consequently, many imaging and sensing applications including microscopy and optical tweezing have turned to using digital holography. A significant obstacle for digital holography in real-time applications, such as wavefront sensing for high energy laser systems and high speed imaging for target racking, is the fact that digital holography is computationally intensive; it requires iterative virtual wavefront propagation and hill-climbing to optimize some sharpness criteria. It has been shown recently that minimum-variance wavefront prediction can be integrated with digital holography and image sharpening to reduce significantly large number of costly sharpening iterations required to achieve near-optimal wavefront correction. This paper demonstrates further gains in computational efficiency with localized sharpening in conjunction with predictive dynamic digital holography for real-time applications. The method optimizes sharpness of local regions in a detector plane by parallel independent wavefront correction on reduced-dimension subspaces of the complex field in a spectral plane.
Digital holography has received recent attention for many imaging and sensing applications, including imaging through turbulent and turbid media, adaptive optics, three dimensional projective display technology and optical tweezing. A significant obstacle for digital holography in real-time applications, such as wavefront sensing for high energy laser systems and high speed imaging for target tracking, is the fact that digital holography is computationally intensive; it requires iterative virtual wavefront propagation and hill-climbing to optimize some sharpness criteria. This paper demonstrates real-time methods for digital holography based on approaches developed recently at UCLA for optimal and adaptive identification, prediction, and control of optical wavefronts. The methods presented integrate minimum variance wavefront prediction into digital holography schemes to short-circuit the computationally intensive algorithms for iterative propagation of virtual wavefronts and hill climbing for sharpness optimization.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.