Open this dialog using the 
button on the main image toolbar.
Create Synthetic Image
The Create Synthetic Image command creates realistic images containing noise and optionally, stars. These images are useful all sorts of analyses like testing algorithms and exploring the effects of noise and noise propagation through image processing. All forms of image noise may be included. Stars have a Gaussianpoint-spread function and follow a realistic stellar luminosity function with randomized intensity and placement. The parameters are specified in commonly used units.
Open this dialog using the button on the main image toolbar.
|
Profile |
Selects a parameter collection for display in the dialog. |
|
Columns |
The number of columns in the new image. |
|
Rows |
The number of rows in the new image. |
|
Data Type |
Select the data type of the new image. |
|
Gain |
The electronic gain of the camera system, in units of electrons per number (also called e/ADU). |
|
Read Noise |
The camera readout noise in units of electrons. |
|
Value |
The background value of the image. This is specified in units of DN (digital numbers). The DN value is related to the number of electrons using the Gain. |
|
Base Noise |
The base level of noise measured in units of electrons. This is a supplemental, additive noise term and usually would be set to 0. |
|
Number |
The number of stars to create. A consequence of the stellar luminosity function is that most stars will be very faint. |
|
FWHM |
The Full Width at Half Maximum brightness, measured in units of pixels. Note: this is a diameter, not a radius. |
|
Max Signal |
The maximum signal used for creating stars. In combination with the Data type and Value parameters, it is possible to set this to a value such that the few brightest stars may be saturated above the counting limit of the image. For example, using Data type = 16 bit unsigned, Value=10000, and Max Signal = 65000 will allow the brightest star to peak at around 75000 DN, which is above the 65535 DN limit of the 16-bit data type. |
|
Number of images |
This specifies the number of independent images to create using the parameter set. If more than 1, the "Open as Image Set" checkbox appears. |
|
Results |
Specifies the output location for listing the properties of the synthetic stars that are created. The options are None (no data are listed), a Mira Text Editor window, and a Mira Report window. Saving the results to a Report window provides added benefits, including the ability to perform analyses on the grid data and to send the image cursor to any star location to identify it on the image (see below). |
|
Open as Image Set |
If Number of images > 1, this specifies whether the images are opened in separate windows or as an image set in a single window. |
Schematically, the image value at any pixel location is given as follows:
Image value = (Background Value + PSF Values + Noise) / Gain.
where the Noise value is a Gaussian random value and may be positive or negative. The PSF value is the sum of all stellar PSF's evaluated at the pixel location. All values are calculated in fundamental units of electrons (or photons) and then converted to IMage Value in units of DN (Digital numbers, or ADU) as would be recorded in an image.
The Noise value above is computed from all sources of which all are assumed to be uncorrelated. The noise can be described as follows:
Noise squared = Sum of Squares of [Background Noise, Shot noise, Base (Readout) Noise, and Digitization noise].
Background noise is the shot noise of the background luminance. Shot noise refers to photon noise in all star PSF's at each pixel location. All noise sources are computed in fundamental units of electrons (or photons).
The following example illustrates the process of creating and analyzing a synthetic star field. The image below shows a portion of the synthetic image produced using the parameters shown in the dialog above. This example created 10,000 artificial stars, mopst of which are faint becuase of the stellar luminosity function.
Data for the 10,000 artificial stars were listed in the Report window shown below. The listed intensity is not the peak value but, rather, gives the total volume under the Point Spread Function ("PSF"). This intensity is the target value for which the PSF was created, and hence it is a "pre noise" value. The actual intensity measured for any articial star will differ because of random noise sources invluded in the model.
The Report window includes a pop-up menu which is opened by right-clicking the mouse, as shown below. After selecting (highlighting) the target star, the "Go to Object" command was clicked to center the image cursor on star 9889 as shown in the Report. The centered star is shown in the image window below.
In the next picture, the Scatter Plot command was selected from the menu. If the positions of the artificial stars are random, then there should be no correlation between x and y coordinates. To investigate this, the scatter plot below was created from the tabular star data. After the plot was created, the Plot Series Attributes command was used to change the default squares to smaller dots:
The following window shows a radial profile for one a medium-bright artificial stars. The calculated FWHM is 2.519, identical within the noise to the 2.5 target FWHM specified in the Create Synthetic Image dialog.
Another test shows the histogram of a region of "sky" background. The background contains many very faint stars which are all positive (though some pixels may lie below the mean background level because of noise in the background signal.
The positive nature of the stars gives a slightly longer tail on the positive side of the histogram median, as can be seen below. Note that the sky histogram has a gGaussian shape beucase of the Gaussian random noise in the model and is centered on 1000 as specified in the parameter dialog.
