Media Cybernetics
From Images to Answers  

Introduction to Deconvolution

Why Deconvolve?
dot How Does Deconvolution Work?
dot Point Spread Function (PSF)
dot Blind Deconvolution
bullet Spherical Aberration Correction
bullet Deconvolution Algorithms


Product Overview
dot NEW in AutoQuant X3
dot AutoQuant Advantage
Compare Versions
System Requirements
bullet Supported File Formats


bullet Acquisitions Dos and Donts
bullet Validating Your Results
bullet Application Notes
bullet AutoQuant Bibliography
bullet AutoQuant Video Tutorials

AutoQuant Brochure PDF

Home > Products > Image Analysis Software > AutoQuant > Deconvolution Algorithms 

AutoQuant Deconvolution Algorithms

No/Nearest Neighbor algorithms work by deblurring one 2D image slice at time. They utilize a subtractive approach based on the simplifying approximation that the out-of-focus contribution in the image slice is equal to a blurred version of the collected adjacent slices. These algorithms are fast, qualitative and work particularly well on images with strong signal to noise ratios.

Inverse Filter or Wiener filter is a one-step image process performed in Fourier space by dividing the captured image by the PSF. This algorithm is a fast and effective way to remove the majority of the blur from widefield images using a symmetric or spherically-aberrated theoretical or acquired point spread function. Image noise is managed through an adjustable smoothing operation applied during processing. Algorithm results are qualitative and generally better than the no/nearest neighbor algorithms especially in the XZ and YZ perspectives.

Non-Blind Deconvolution is a constrained iterative approach that requires a measured or synthetically acquired PSF for processing. This algorithm is based on the same statistical and computational foundation of AutoQuant's renowned Adaptive Blind algorithm and shares the same superior noise handling characteristics and flexibility. However, the PSF provided is assumed to be accurate and is not modified during the deconvolution. Non-Blind offers an excellent balance between quality results, quantitative analysis and time to process.

Adaptive Blind AutoQuant's Blind Deconvolution algorithm draws upon the statistical techniques of Maximum Likelihood Estimation (MLE) and Constrained Iteration (CI) to produce the most robust and statistically accurate results available on the market today. It does not require a measured or acquired PSF, but instead iteratively reconstructs both the underlying PSF and best image solution possible from the collected 3D dataset. It is well suited for environments where signal to noise ratios are challenging and operates across the full spectrum of modalities.

2D Blind Deconvolution is an adaptive method for 2D data that does not require your microscope and image parameters. 2D Blind Deconvolution works by iteratively improving the data set and works with time series image sets, individual color channels or intensity images. 2D Blind Deconvolution is capable of restoring features at a sub-pixel resolution level and can work with almost any 2D image.

2D Real-time uses AutoQuant's powerful 2D blind deconvolution algorithm to remove blur from a single image. No microscope or image parameters are required. Useful Sharper/Smoother, thickner/thinner and brighter/dimmer controls help guide you to get the very most out of your data. Deblur one frame instantaneously or several in near real-time.

Register Products    Contact Us    Site Map


Copyright 2012 Media Cybernetics, Inc. All rights reserved.