HOME > Support >ManualContext >
Noise reduction for FPN noise(You must read this before capturing Deep Sky Objec
2018 年 01 月 26 日
Chapter7: The theory of FPN Noise and Calibration Method 
You must read this chapter to get best image of DSO
 
 
 
 
Theory of miniCAM5F's FPN Noise
 
The theory of CMOS and CCD has big difference. It will cause the different noise calibration method. CMOS sensor has build-in ADC convertor. The modern CMOS has column ADC. Each column has an ADC. It is very easy to get high frame rate and low readout noise. Because all ADC is working in parallel. Each ADC can get low readout speed.
 
But this structure brings the FPN noise. Each ADC have small different response and those differences cause each colmn is not the same. Many vertical lines appear on the images. 
 
To reduce the FPN noise. The modern CMOS sensor has an on-chip FPN calibration circuit. This circuit can meassure the FPN noise once and calibrate it. It can get rid of the FPN noise but  still a little FPN noise will remain. When you make a high stretch ratio you will see it. In order to keep the background clean and vertical lines free, you must do the FPN calibration.
 
 
Let's introduce some theory here. The following formulas show all signals in an image we get from miniCAM5F.
 
 
Light Frame = Actual light + Dark current + bias offset + FPN noise (1)
Dark Frame =                        Dark current + bias offset + FPN noise (2)
Bias Frame =                                                  bias offset + FPN noise (2)
 
What we want to get is the Actual light. This represent the light come onto the sensor.
 
We can use the subtract method to do that. 
 
 
Light Frame - Dark Frame = [ Actual light + Dark current + bias offset + FPN noise (1) ] - [ Dark current + bias offset + FPN noise (2) ]
                                              =   Actual light  + FPN noise(1)- FPN noise(2)
 
 
If the FPN noise (1) = FPN noise (2) , the results is the Actual light. You can get perfect image. For CCDs this is no problem.
 
But miniCAM5F has an internal FPN calibration circuit. After its calibration, it gets rid of most FPN noise but still some remain. The remain  FPN noise may be different for light frame and dark frame.  Because each time when CMOS sensor reset , camera bootup , or gain changed, the CMOS sensor may do a calibration. It causes the remain FPN noise is not the same.
 
This means you can not use directly subtract method to get calibration, except you capture the light frame and dark frame at the same time and there is not reboot , reconnect or software restarting happen.
 
 
Also please note the FPN noise has relationship with temperature. You must capture the light frame and dark frame at the same temperature. 
 
 
 
 
How to get perfect FPN noise calibration?  We will introduce the method in following text.
 
 
 
FPN Calibration Method for miniCAM5F
 
 
Two method can be used in actual imaging. You can select the method you prefer. Please note when the FPN noise is calibrated, the dark noise is also calibrated. So you can use the processed image to stacking directly without do dark frame calibration anymore.
 
 
Method 1   Light Frame - Dark Frame
 
The formula is
 
 
 
Light Frame - Dark Frame = [ Actual light + Dark current + bias offset + FPN noise (1) ] - [ Dark current + bias offset + FPN noise (2) ]
                                              =   Actual light  + FPN noise(1)- FPN noise(2)
 
 
In this way you must make sure the dark frame and light frame is captured at one time without any accident happen. Like the camera reboot or software restart etc. So that the FPN noise(1)- FPN noise(2) = 0  
 
 
 
 
 
 
 
Operation Steps
 
1. set target temperature and wait camera get it
2. cover the telescope or lens . Capture some dark frames
3. open the telescope or lens. Capture light frame
4. average all dark frames to get a master dark frame.
5. calibrate each light frame with master dark frame. It is best to add a constant (Pedestal) to prevent some pixel is below zero.
6. stack the light frames.
 
This way is simple but you must take the dark frame each time. If the exposure time is long  you may waste a lot time to capture it.
For miniCAM5F, it comes with a black filter. You can control the color wheel to rotate to this position to capture dark frame. You can do the automatic dark frames capture in Planner Table.
 
 
 
 
 
 
Method 2: (light frame - bias frame1)-  (dark frame -bias frame2)
 
 
The formula is 
 
 
 
Light Frame 
 
 
Light Frame  - Bias Frame1  =  [ Actual light + Dark current + bias offset + FPN noise (1) ] - [ bias offset + FPN noise (2) ]
 
if the light frame and bias frame are captured at the same time, the results is     Actual light + dark current 
 
Dark frame - Bias Frame2 =   [  Dark current + bias offset + FPN noise (3) ] - [ bias offset + FPN noise (4) ]
 
if the dark frame and bias frame captured at the same time, the results is     dark current 
 
Finally 
 (light frame - bias frame1)-  (dark frame -bias frame2) = Actual light
 
 
Since bias frame exposure time is zero. It is very fast to capture it. You can capture the bias frames each time when you begin a planner. Then you can use your dark frame library to calibrate that.
 
 
 
 
 
 
 
Operation Steps
 
1.Make Dark Frame library
 
1.1. Set the target temperature and wait the camera get it
1.2. cover the telescope or lens. Capture some dark frame. You can make dark frame for each exposure time you are using. like 30sec,1minute,2.....
1.3. cover the telescope or lens, Capture some bias frame.
1.4. average the dark frames to get Master dark frame. average the bias frame to get the Master Bias average frame.
1.5. Master Dark frame subtract Master Bias frame to get the "Master dark frame with FPN noise free" It is best to add some constant (Pedestal) .Save it .Such Master dark frame can be used at anytime( at the same sensor temperature)
 
 
2.Calibration
 
2.1. Set the target temperature and wait the camera get it.
2.2 Cover the telescope or lens. Capture some bias frame (exposure is zero)
2.3. Open the cover and capture light frame.
2.4. average the bias frame to get the Master bias frame.
2.5. Subtract the Light frame with the Master bias frame to get the "Light frame with FPN noise free" 
2.6. Stack all light frame.
 
With this method, you does not captue dark frame each time. Only need to capture bias 
Newsletter
         
Meet us
Contact us

  503, Block A, Singularity Center, Shahe Town, Changping District, Beijing, China, 102206

  +86(10)-80709022-602

  sales@qhyccd.com        support@qhyccd.com

     

Copyright  2011-2018  QHYCCD    Question  |  Navigation  |  Forum