Camera application-dependent factors such as Fixed-Pattern Noise (FPN), Photo Response Non-Uniformity (PRNU), Illumination and Optical non-uniformities will have a negative impact on the quality of images captured. The effects of these parameters can be minimized and often eliminated by performing a proper Flat-Field Calibration (FFC).
FFC data is stored in non-volatile memory inside the camera. All FFC-enabled Pixelink cameras are shipped with factory calibrated FFC, done at a gain of 0dB and at an integration time of 40ms – 100ms depending on the camera model, without a lens under very uniform light using an integrating sphere. Since the camera will be operated in your application environment, which will have its own unique set of operating conditions (lighting, optics, gain, integration time), it is imperative that you perform the Flat-Field Calibration using your operating environment.
Regardless of the type of light source used it is extremely difficult, if not impossible, to have a perfectly even illumination pattern on the object being imaged.
Without FFC – Non-uniform Light Source Following FFC
The above images were taken against a flat white board using a single light source positioned to the left of the camera. Notice the effect that the light source has before FFC was performed. After FFC has been completed, the same white board appears evenly illuminated using the same camera settings, camera lens and light source.
Fixed-Pattern Noise (FPN)
Fixed Pattern Noise (FPN) is a fixed pixel-to-pixel offset and is due to a combination of variations in image pixel geometry, substrate material and dark current. It is fixed with respect to illumination; however, it is temperature, integration time and gain dependent. FPN is measured under dark conditions at a given integration time, temperature and gain.
Dark current is linear with respect to integration time and doubles with every 6-8 degree Celcius increase in temperature. As such, a camera which has been factory calibrated and is then used in a high-heat environment will exhibit much more dark current than when it was originally calibrated. Performing FFC in the operating environment will compensate for this.
Illustration of Fixed Pattern Noise in a 16 Pixel (4x4) Area Array (Magnified)
Photo Response Non-Uniformity (PRNU)
Theoretically, when a fixed, uniform light falls on the pixels of a CCD/CMOS sensor, each cell should output the exact same voltage. However, due to a variety of factors including slight variations in pixel geometry and substrate material, this is not the case. When a uniform light is transmitted onto the pixels of a CCD or CMOS camera, the pixels output slightly different voltages. This difference in response to a uniform light source is referred to as Photo Response Non-Uniformity or PRNU. Since PRNU is caused by the physical properties of a sensor, it is nearly impossible to eliminate and is therefore usually considered to be a normal characteristic of the sensor array used in any CCD or CMOS camera.
The following images show the effects of PRNU and how it becomes more pronounced with increased integration time and gain. The first image was taken using a 38ms integration time with 0dB gain. The second image was taken using a 240ms integration time with 18.1dB gain and the third image also uses a 240ms integration time with 18.1dB gain, but with a substantially improved image due to FFC. Histograms for each image are also provided.
38ms integration, 0dB gain before FFC
240ms integration, 18.1dB gain before FFC
240ms integration, 18.1dB gain after FFC
Several factors must be considered when determining the quality of a lens including sharpness, contrast, color correction, relative illumination, spectral transmission and distortion. These factors result in lenses producing slight variations between the object and the image.
Lens intensity varies from the center to the edge of the image with the center being brighter than the edges. This can often affect the suitability of a lens for your application and should be examined prior to integrating a lens into your application. This non-uniformity is caused by two factors: the first is the natural decrease in illumination from the center of the image circle to the edge of the lens and is known as the cosine fourth effect. The second, known as mechanical vignetting, is often present and results in some rays being blocked by mechanical parts located within the lens. Vignetting of this type can often be greatly reduced by using a smaller aperture.
Using the Histogram to Implement Even Illumination
The histogram feature provided in Pixelink Capture OEM can be useful in configuring a light source to optimize the illumination pattern to be as uniform as possible. To access the histogram, click View | Histogram or press Ctrl+H while in Pixelink Capture OEM. A uniform light source is represented graphically by a very narrow or “tight” histogram, whereas non-uniform lighting is shown as a broad or wide histogram. The following histograms and their respective images, illustrate uniform and non-uniform light. To use the histogram as an aid in configuring your light source, enable the histogram, adjust your camera settings and then adjust your light source until the histogram is as narrow as possible.
Uniform Light (Mono Camera) Non-Uniform Light (Mono Camera)
Uniform Light (Color Camera) Non-Uniform Light (Color Camera)
Uniform Light Image Sample Non-Uniform Light Image Sample
Before Performing Flat-Field Correction (FFC)
Ensure that you have a flat white or grey card that covers in the entire Field of View (FOV) of the camera before proceeding with the FFC Wizard. If the lens is focused perfectly, the camera will be able to image the surface irregularities (texture) of the card and the FFC will compensate or correct for the surface texture. Therefore it is important to defocus the lens when capturing the FFC bright field image (see step #7 below).
For optimum image quality, the FFC wizard supplied by Pixelink should be used to perform Flat Field Calibration for each camera in its specific application environment. Changes in lighting, optics or camera settings after the FFC has been performed may result in the camera not delivering optimum image quality. If any of these variables are altered, FFC should be performed again.
Before performing FFC, the bright field image mean must be between 55% & 80% of Saturation level. If it falls outside of this range the FFC wizard will not work properly. Image mean can be adjusted using the iris on the lens, however this is not always an acceptable or available method of adjusting this property. As such, we recommend that image mean be adjusted to an acceptable value using the camera settings (gain and/or integration time) prior to running the FFC wizard. 55% & 80% translate into 114.75DN & 178.5DN on the histogram respectively. The following images illustrate acceptable and unacceptable bright field image mean as well as how the information is shown on the histogram. The mean value below the histogram is the value that we are concerned with and it should be between 115DN & 179 DN.
Acceptable Bright Field Image Mean & Corresponding Histogram
Unacceptable Bright Field Image Mean & Corresponding Histogram (Too Low)
FFC Wizard Steps
- From within Pixelink Capture OEM, set all camera settings to the values that are to be used in the application environment. These settings include, Exposure Time, Gain, Brightness and Pixel Format. Also, the Region of Interest should be set to the maximum camera resolution.
- After the camera settings have been set, click Camera | Save Camera Settings and then save the configuration to Channel 1.
- After the camera settings have been saved, exit Pixelink Capture OEM and launch the FFC Calibration Tool by clicking Start | Programs | Pixelink | Tools | FFC Wizard. When the FFC Calibration Tool opens, review the pre-calibration notes and click Next to continue
- You will have the choice to either Calibrate FFC or Load FFC From File. The later is useful if you must revert back to a previously saved configuration for any reason. Click Calibrate FFC and click Next.
- It is strongly advised that you save the existing FFC configuration prior to proceeding to the next step. In the event that the FFC generates undesired results or fails, you will then be able to restore the FFC calibration to its factory default state. If you do not want to save the existing configuration, enable the check box labelled Skip This Step and click Next.
- The FFC Wizard is now ready to capture Dark Fields. Before clicking next to begin capturing Dark Fields, ensure that the lens has been capped with a lens cap that does not permit any light from entering the camera. If possible, turn off or block any light including ambient light sources as even the slightest amount of light can generate undesired results if allowed to enter the image sensor.
- After Dark Field images have been captured, the FFC Wizard will be ready to capture Bright Fields. Remove the lens cap from the camera and ensure that the light source and optics used in your application have been set to the normal operating conditions. Click Next to continue.
- In the event that the Image Mean is not between 55% and 80%, you will be prompted with an error similar to the following:
The image mean can be altered either by adjusting the lens iris so that it is more open or closed (depending on whether Image Mean is below 55% or above 80%) or by adjusting the camera settings. If you cannot adjust your iris to correct this, you must Abort the FFC Wizard and then adjust your camera settings using Pixelink Capture OEM so that Image Mean falls between 55% & 80% saturation. You can determine when the Image Mean is acceptable by using the Histogram as illustrated at the beginning of the Performing FFC section.
- After Bright Fields have been captured, the FFC Calibration will have to be saved to the camera. Click Next to begin writing the data to the camera.
- When the data has been written, click Finish to exit the FFC Wizard.
- Disconnect and reconnect your camera for the new FFC Calibration to take effect.