DBE |
|
Dynamic background Extraction |
|
Removing gradients and uneven illumination is key to getting good results with your images If there is one tool worth getting to know well this is it It works by taking samples of the background and creating a background model and then normally subtracting it from your image |
|
To start with we will have a look at some of the variablesThis is the panel of the tool and 2 drop panels opened, these will deal with placement and interpretation of the samples |
|
Selected SampleA) Sample number and position B) Radius Size of current sample C) RGB weights D) Sample navigation E) Content of selected sample Model ParametersF) Tolerance -- high setting inc more pixels in the model, but with the risk that they might not relate to the true background, default o.5 is a good place to start , this is a setting that has a important place and we will look at this shortly G) Shadows relaxation: Increasing this parameter allows for the inclusion of more dark pixels in the generated background model, while more restrictive criteria can be applied to reject bright pixels (as specified by the tolerance parameter). This helps to achieve a better sky background modelization without inclusion of relatively bright and large objects, as extended nebulae---Suggest leave as default . H) Soothing factor -- to do with the way the background model is created , suggest left at default for normal gradient removal , but try lower numbers if you have errors that change over small scales e.g flat problems
Sample generationI) Default sample radius-- this is the sample size that will used---set to suit your image i use around 14 on a 6 meg pix image , suggested range from 4 to 20 J) Samples per row ---Set to suit your image , see examples k) Min Sample Weight--Lower number puts on more samples L) Sample settings J
|
Fig 1
|
This is the panel of the tool and the bottom drop panel opened which will generate and apply the background model |
|
Target Image Correction
N) Correction--This is how we apply the model to the image , for virtually all purposes we will subtract it O) Normalize --If this option is selected, the initial median value of the image will be applied after background model correction. If the background model is subtracted, then the median will be added; if the background model is applied by division, the resulting image will be multiplied by the median. This tends to recover the initial color balance of the background in the corrected image. I tend not to use this suggest leave as default i.e. unchecked P) I personally leave this unchecked ( Default setting ) as I like to see the background model , can help with sample positions if correction is not correct Q) Again I personally leave this default ( Unchecked ) so I can reapply DBE to the original image if I am not satisfied with first iteration R) Suggest leave as default so output image is same as original |
Fig 2
|
How to use DBE |
|
We break down the application of DBE into three areas 1) Placing of samples to the image 2) examine the quality of the samples and adjusting setting if required 3) Application of the background model |
|
I will start by looking at a fairly easy image with lots of free background area |
|
Placing the samples1) Open your image and apply a strong STF to see where there is free background ( hopefully) 2) Click on the image to activate DBE ( the image will be divided into four ) 3) We now need to place some samples on the image on free background areas , you can do this one at a time with your mouse and clicking on the screen or use the sample generation part of DBE ( I to M ) and this is we will do next by doing 4) Select a sample radius for your sample and use the goldilocks theorem here not to small and not to large I have used 14 ( I ) 5) Set samples per row ( J )and for this image use the same rule as 4 , but its not always about the amount of samples it is the quality to 6)Sample setting ( K ) try applying the samples to the image ( by clicking the generate button M ) with the default setting first and see if the required amount of samples is applied if not alter ,I have used 0.15 below, quite low as the image I am using has had no flat applied so there is a large illumination difference 7) If you need to add sample just click on the image to add them |
|
Your image should start to look like thisClick image to enlarge fig 3 |
|
 |
|
Sample quality and settings
1) Bear in mind While viewing the sample contents box ( E ) the following 2) Now we have to check that the samples are not placed over areas of interest , in this case the galaxy if there are either delete them by selecting them with your mouse and hitting the delete key or by moving them with your mouse 3) Another important point is to make sure samples are not place near bright stars ( do not worry about fainter ones ) as they can influence the background model 4) Now take a closer look at the image above (fig 3) above and you will see that some of the samples are coloured red this means there is not enough weight inc in that sample to form a correct background model also if you look in the sample box ( Mouse has selected one of the red samples ) you will see it is very dark , they need to look like the image below ( FIG 4 ) , notice how the samples are now not red and the contents are light in colour 5) How did I achieve light coloured samples , If you look at the tolerance I have increased the tolerance to 0.8 this is not a risk in this image as I know it is clear background and I will not be including any nebula in the background model , just click on the image after altering a parameter for the settings to update the image with the new settings
|
|
| Click on image to enlarge Fig 4 | |
 |
|
Application of the model |
|
1) Now go to the bottom panel ( N to R ) and select subtraction under Correction 2) Leave the other settings as default 3) apply the process to the image After this you should have some thing like the image below ( Fig 5 ) , please note the output images will be linear so I have applied a STF to both the background model and the new image A good correction seems to have been carried out , but if this has not happened you may need to shift / delete some samples if holes ( dark areas ) have appeared in your image |
|
| Click on image to enlarge Fig 5 | |
 |
|
Click here for a more difficult example with lots of nebula |
|
