Once instrumental signatures are removed from the raw data in the initial reduction process (see Data manipulation). We are ready to derive scientific results out of them. But we can’t do anything special with a raw dataset, for example an image is just an array of values. Every pixel just has one value and its position within the image. Therefore, the first step of your high-level analysis will be to classify/label the dataset elements/pixels into two classes: signal and noise. This process is formally known as detection. Afterwards, you want to separate the detections into multiple components (for example when two detected regions aren’t touching, they should be treated independently as two distant galaxies for example). This higher level classification of the detections is known as segmentation. NoiseChisel is Gnuastro’s program for detection and segmentation.
NoiseChisel works based on a new noise-based approach to signal detection and was introduced to the astronomical community in Akhlaghi and Ichikawa . NoiseChisel’s primary output is an array (image) with the same size as the input but containing labels: those pixels with a label of 0 are noise/sky while those pixels with labels larger than 0 are detections (separate segments will be given positive integers, starting from 1). For more on NoiseChisel’s particular output format and its benefits (especially in conjunction with MakeCatalog), please see Akhlaghi . The published paper cannot under go any updates, but the NoiseChisel software has evolved, you can see the major changes in NoiseChisel changes after publication.
Data is inherently mixed with noise: only mock/simulated datasets are free of noise. So this process of separating signal from noise is not trivial. In particular, most scientifically interesting astronomical targets are faint, can have a large variety of morphologies along with a large distribution in brightness and size which are all drowned in a ocean of noise. So detection is a uniquely vital aspect of any scientific work and even more so for astronomical research. This is such a fundamental step that designing of NoiseChisel was the primary motivation behind creating Gnuastro: the first generation of Gnuastro’s programs were all first part of what later became NoiseChisel, afterwards they spinned-off into separate programs.
The name of NoiseChisel is derived from the first thing it does after thresholding the dataset: to erode it. In mathematical morphology, erosion on pixels can be pictured as carving off boundary pixels. Hence, what NoiseChisel does is similar to what a wood chisel or stone chisel do. It is just not a hardware, but a software. In fact looking at it as a chisel and your dataset as a solid cube of rock will greatly help in best using it: with NoiseChisel you literally carve the galaxies/stars/comets out of the noise. Try running it with the --checkdetection option to see each step of the carving process on your input dataset. You can then change a specific option to carve out your signal out of the noise more successfully.
|• NoiseChisel changes after publication:||Changes to the software after publication.|
|• Invoking astnoisechisel:||Options and arguments for NoiseChisel.|