KB Imaging Solutions LLC Image processing R&D
Get in touch
KB Imaging Solutions LLC

Getting more out of the image you already have

Image restoration, deconvolution and measurement software for scientific, industrial and consumer imaging — plus consulting and contract work in digital signal and image processing.

Founded 2006. Independent, privately held, and still writing the code.

See the work Start a conversation
✓ Blind deconvolution and image restoration ✓ Published, widely cited algorithms ✓ Microscopy, astronomy and instrument imaging ✓ Consulting and contract development
What we do

Image processing, from the mathematics up

Most of the work here starts with the same question: what did the optics, the sensor and the atmosphere do to this image, and how much of it can be undone?

Deconvolution & image restoration

Blind and non-blind deconvolution, point spread function estimation, noise modelling and iterative restoration — the core research area of the company, and the subject of published work that is still cited and shipped in commercial software today.

Scientific & industrial imaging software

Complete applications rather than prototypes: acquisition, calibration, registration, restoration, measurement and export, built to run on real data on ordinary hardware, and to be used by people who are not image processing specialists.

Instrument & device integration

Getting the data off the hardware in the first place: camera and instrument drivers, serial and SDK-level device adapters, acquisition pipelines and control software for laboratory instruments and scientific cameras.

Consulting & contract programming

Available for algorithm development, feasibility studies, second opinions on an approach that is not converging, and contract implementation in MATLAB, C++ and Java. Short engagements are fine; so are long ones.

Selected work

Some of what has come out of it

A short record rather than a full one — the projects that best show what the company does.

  • Current Public beta

    BiggSky — astrophotography image processing

    A complete processing application for astrophotography: calibration, frame grading, stacking, blind deconvolution, colourisation and export, all in one program. The deconvolution measures the blur from the data itself, so it works on the lunar surface, a solar disc or a planet — subjects with no usable reference star anywhere in frame. Windows, macOS and Linux.

    biggsky.com
  • Ongoing

    Instrument software and device integration

    Acquisition and control software sitting between scientific hardware and the people using it: serial and SDK-level drivers for scientific cameras, device adapters for open-source microscope control platforms, and analysis code for impedance-based cell measurement instruments. Cross-platform C++, Java and MATLAB.

  • Earlier

    3D deconvolution for optical microscopy

    Blind deconvolution for widefield, confocal and multi-photon microscopy — the AutoDeblur and AutoQuant line of products, where adaptive PSF estimation was taken from a research result to software that microscopists used daily without needing to know what a point spread function was.

  • 1997 Still in use

    Accelerated iterative restoration

    A vector-extrapolation method that cuts the number of iterations a Richardson–Lucy style restoration needs, typically by an order of magnitude, without changing the answer it converges to. It is implemented in MATLAB's deconvlucy and deconvblind, and in a good deal of scientific imaging software since.

    The paper
Current project

BiggSky

The same restoration mathematics, pointed at the night sky.

BiggSky takes raw frames — long-exposure deep sky stacks, or SER and AVI captures of the Moon, Sun and planets — and carries them through calibration, stacking, deconvolution, colour and export without leaving the program. Nothing is invented and nothing is hallucinated: what comes out was recorded by the sensor, it was just buried under the blur.

  • Blind deconvolution — no reference star, no radius to guess
  • Tiled solving, so seeing that varies across a wide field is handled
  • Runs on your own machine; nothing is uploaded anywhere
  • Windows, macOS and Linux
Visit biggsky.com

A new biggsky.com is on the way — downloads, gallery, tutorials and licensing all in one place.

Public beta BiggSky

Calibrate. Stack. Enhance. Publish.

Research

The published work behind it

The company grew out of research on iterative image restoration. Some of that work is still running inside other people's software.

  • Acceleration of iterative image restoration algorithms D. S. C. Biggs and M. Andrews. Applied Optics, vol. 36, no. 8, pp. 1766–1775, 1997. The vector-extrapolation scheme used by MATLAB's deconvlucy and deconvblind.
  • Accelerated iterative blind deconvolution D. S. C. Biggs. PhD thesis, University of Auckland, 1998. Where the acceleration and the blind PSF estimation were worked out in full. Read it on ResearchSpace
  • Adaptive blind deconvolution and PSF estimation Subsequent work on estimating the point spread function directly from the data, and on applying it to three-dimensional microscopy volumes — the basis of the AutoDeblur and AutoQuant deconvolution products, and today of BiggSky.
  • Full publication record The complete list, with citations, lives on Google Scholar and ResearchGate.
Working on a restoration problem that will not behave? Deconvolution fails in characteristic ways — ringing, noise amplification, a PSF estimate that collapses. Those failures are usually diagnosable. Send over the details.

Let us know what we can do for you

Consulting, contract development, or a question about one of the projects above — email reaches a person, and gets a reply.

info@kbimaging.com LinkedIn