January 2015
Spotlight Summary by Richard Bowman
Digital holography super-resolution for accurate three-dimensional reconstruction of particle holograms
In-line digital holography is fast becoming a widespread method for imaging and tracking micro-objects in three dimensions. Employing a simple bright-field microscope and adding transmission illumination from a collimated laser diode, it is possible to form interferograms, essentially the diffraction pattern of the micro-object being imaged. These interferograms allow the reconstruction of objects in three dimensions. A common approach when tracking particles with this technique is to use an analytic model for the particle’s diffraction pattern - this is possible with many commonly-used particles, such as spherical colloids. Each frame in a video sequence can be used to estimate the 3D position of an object, as well as the parameters that describe its interferogram (for example, size and refractive index).
Verrier and Fournier present a method where this tracking can be improved; after an initial rough tracking step, small changes in the object’s position between images can be used to create a super-resolved image of it. The resulting image has both a higher resolution and better signal-to-noise ratio than the individual frames. Fitting the model to this higher-quality image allows them to estimate some parameters (for example axial position and size) more accurately. Once those parameters are established, they can be used to calculate a better template image for the particle, which can then be used to estimate for the particle’s position more accurately in each video frame. This leads to an improvement in the lateral tracking accuracy as well. Absorbing particles were simulated for this paper, the diffraction patterns of which can be parameterised by their radius and axial position. The lateral motion of the particles was used to create enhanced images to find radius and axial position, which in turn led to improved template images for lateral tracking.
The authors suggest their technique should be generalised to include axial position in the super-resolution algorithm as a parameter that can change from frame to frame. In most microscopic situations, this would be optimal: Brownian motion means a micro-object’s position changes quite rapidly in three dimensions, but generally other parameters (like size or refractive index) do not change on short timescales. Using a super-resolution technique to create a high-quality image that allows these constant parameters to be estimated more accurately then means that the particle’s motion in 3D can be tracked better, by comparing each image with the best possible template image.
As the method presented already copes with multiple objects, which may move independently, it should be suitable for a wide range of situations where particles must be tracked as accurately as possible. This should further improve the accuracy with which in-line holography can localise particles, in applications from micro-particle imaging velocimetry to optical tweezers.
You must log in to add comments.
Verrier and Fournier present a method where this tracking can be improved; after an initial rough tracking step, small changes in the object’s position between images can be used to create a super-resolved image of it. The resulting image has both a higher resolution and better signal-to-noise ratio than the individual frames. Fitting the model to this higher-quality image allows them to estimate some parameters (for example axial position and size) more accurately. Once those parameters are established, they can be used to calculate a better template image for the particle, which can then be used to estimate for the particle’s position more accurately in each video frame. This leads to an improvement in the lateral tracking accuracy as well. Absorbing particles were simulated for this paper, the diffraction patterns of which can be parameterised by their radius and axial position. The lateral motion of the particles was used to create enhanced images to find radius and axial position, which in turn led to improved template images for lateral tracking.
The authors suggest their technique should be generalised to include axial position in the super-resolution algorithm as a parameter that can change from frame to frame. In most microscopic situations, this would be optimal: Brownian motion means a micro-object’s position changes quite rapidly in three dimensions, but generally other parameters (like size or refractive index) do not change on short timescales. Using a super-resolution technique to create a high-quality image that allows these constant parameters to be estimated more accurately then means that the particle’s motion in 3D can be tracked better, by comparing each image with the best possible template image.
As the method presented already copes with multiple objects, which may move independently, it should be suitable for a wide range of situations where particles must be tracked as accurately as possible. This should further improve the accuracy with which in-line holography can localise particles, in applications from micro-particle imaging velocimetry to optical tweezers.
Add Comment
You must log in to add comments.
Article Information
Digital holography super-resolution for accurate three-dimensional reconstruction of particle holograms
Nicolas Verrier and Corinne Fournier
Opt. Lett. 40(2) 217-220 (2015) View: Abstract | HTML | PDF