PixInsight — Pleiades Astrophoto

Cover Images: First Previous Next Current

Welcome

Pleiades Astrophoto boasts a young, international research and development team with expertise in astronomy, mathematics and software development. Our mission is to provide cutting edge image processing tools for a broad range of technical imaging applications. We design and implement novel paradigms and innovative methodologies.

Open-Architecture Image
Processing Platform

PixInsight is a modular, open-architecture, portable image processing platform. The core PixInsight application provides powerful graphical, command-line and scripting interfaces, and includes a high-performance processing infrastructure. Processes, file formats, and the definitions of their associated user interfaces, are they all implemented as external modules.

Advanced Development
Framework

PixInsight modules are built around the PixInsight Class Library (PCL). The PCL is a platform-independent, ISO C++ development framework that implements a high-level API for communication between modules and the core PixInsight application. The PCL provides a comprehensive set of image processing algorithms, ranging from geometric image transforms to avant-garde multiscale analysis tools. The PCL is free and publicly available.

Development Highlights

Deconvolution Processes

The Deconvolution tool is our implementation of state-of-the-art regularized Richardson-Lucy and Van Cittert deconvolution algorithms. Simply put, these algorithms separate significant image structures from the noise at each deconvolution iteration. This is possible thanks to specific, wavelet-based multiscale analysis techniques.

At each deconvolution iteration, significant structures are kept and the noise is discarded or attenuated. This allows for simultaneous deconvolution and noise reduction, which leads to robust deconvolution procedures that yield greatly improved results when compared to traditional or less sophisticated implementations. In addition, this tool implements a deringing algorithm to fix the well-known Gibbs phenomenon, or ringing artifacts (e.g., dark halos around stars).

Regularized Van Cittert, Wiener and Constrained Least Squares are excellent deconvolution algorithms for lunar and planetary images. Regularized Richardson-Lucy is the best option to deconvolve linear deep-sky images.

We have authored several fully-worked, step-by-step processing examples that will help you to integrate this powerful tool in your image processing workflow. You'll find them along with more relevant information in the links below.

Deconvolution of a High-Resolution Lunar Image, with original CCD raw data acquired by Vicent Peris.

NGC 5189 from Geminy Observatory South: Deconvolution and HDRWaveletTransform in PixInsight, with original raw data from the Gemini Science Archive.

A post on PixInsight Forum with test images and mouseover comparisons. This is a comparison of the regularized Van Cittert, Wiener, and Constrained Least Squares deconvolution algorithms applied to a high-resolution lunar image.

HDRWaveletTransform Process

HDRWaveletTransform is a novel algorithm for multiscale processing of high dynamic range images. It is a completely new, revolutionary vision of the high dynamic range problem. Created by PTeam's wavelet expert Vicent Peris, HDRWaveletTransform has changed the way we process most deep sky images.

The best part is that this tool is both extremely powerful and very easy to use. You can check this on the following documents:

Introductory step-by-step processing example: Multiscale Processing with HDRWaveletTransform
A Selection of Processing Examples with HDRWaveletTransform

Here is another good example with a high dynamic range luminance image of the M42 region:

Linear combination of exposures from 1 minute to 30 minutes with a modified Canon 350D DSLR @ 400 ISO (6.5 hours total exposure time) and instrument apertures ranging from f/8 to f/2.8. All individual exposures were preprocessed and registered with DeepSkyStacker, and integrated linearly as a 32-bit floating point HDR image with PixelMath in PixInsight. This is the luminance of the original RGB color image.

After HDRWaveletTransform (5 layers, 3x3 Gaussian scaling function, automatic balance).

Images courtesy of Vicent Peris and José Luis Lamadrid. Note that only HDRWaveletTransform has been applied with nearly default settings; no other process has been used to produce the second image linked above. Note also that the HDRWaveletTransform process cannot be applied to the above JPEG image; at least a 32-bit image is necessary to represent the whole original data range.

GREYCstoration Process

GREYCstoration is a new noise reduction process in PixInsight. It is our implementation of an open-source image regularization algorithm created by David Tschumperlé, a CNRS researcher in the Image Team of the GREYC Lab in Caen, France. This algorithm is based on state-of-the-art image processing methods using nonlinear multi-valued diffusion partial differential equations.

As implemented in PixInsight, GREYCstoration is a high-performance, all-terrain noise reduction tool, able to preserve extremely thin image details, and adaptable to numerous noise types and image restoration requirements. Our implementation is focused on the denoising capabilities of the algorithm; future versions will also cover inpainting and upsampling with GREYCstoration.

GREYCstoration website
GREYCstoration demonstration examples

PixInsight and the ALHAMBRA Survey

A team of researchers from institutes all over the world, led by Mariano Moles (Instituto de Astrofísica de Andalucía, CSIC), is currently undertaking a large scale survey at Calar Alto Observatory that will reveal the 90% of the history of the Universe. PixInsight is being used to represent ALHAMBRA fields as they would display if observed by the human eye.

Official press release at Calar Alto Observatory website.
A high-resolution image of ALHAMBRA field F08_P01_1, generated with PixInsight.
ALHAMBRA Survey server