Dark-frame subtraction

Image sensor calibration is often a two-stage correction process of the raw image of an image sensor using a white image and a dark frame.

Dark image

A dark image ( engl. Dark Frame ) is a recording of a covered CCD or CMOS camera with the same exposure time and operating temperature as the raw image. It is used especially for long exposures (eg in astrophotography ) to first remove the emerged from the dark current noise. To this end, the dark image is subtracted from the raw image ( dark image or Dark -frame correction). The individual pixels react differently to temperature effects, that is, generated for a given CCD temperature of each pixel has a different power for black. For a particular pixel, the dark current varies by a constant value, so that correction is possible. Pixels with very high dark current, which thus appear bright in the image, hot pixels are called. In scientific applications such as spectroscopy, the dark current correction is often necessary to minimize the detection limits for materials. In particular, for processes such as DOAS the dark current correction is often imperative as the ' dark current spectra ' often exhibit strong differential structures that can falsify the measurement results significantly. Even in astronomy, this procedure is common.

White image

To compensate for differences in brightness in the image, which go back to impurities on the sensor (dust), non-uniform sensitivity of the pixels, or the optics used ( vignetting, reflexes ), the white image is recorded by the instrument is aligned on a uniformly lit surface. The raw image (from which the dark image has been subtracted) is indicated by the white image divided (white screen or flat-field correction) and multiplied by the average of the white image. For example, if a pixel of the chip is covered by dust to 50%, the measured intensity is reduced there, both the image of the captured object and the white image. Due to the division by the small level of brightness in white image, the brightness value is raised image of the object to the value that he would have no dust.

Example

The images show the calibration using the example of an astronomical recording:

  • The raw image shows several "hot pixels" that lead to a very noisy image. Faint stars are lost in this noise. Dark spots in the raw image are due to dust in the CCD camera here.
  • The dark image was at the same exposure time and operating temperature as the raw image, but with the camera shutter closed, was added. It thus captures the interest accrued during the corresponding exposure time dark current.
  • The white picture was taken while the instrument was aligned on an evenly lit surface. It detects irregularities in the illumination of the image ( for example, by dust ) and the sensitivity of each pixel.
  • When calibrated image these defects are corrected. Even faint stars are visible here. The calibrated image can be used for quantitative analysis, for example for measurement of the apparent brightness of stars used. Without calibration, such a measurement would lead to false readings.

Dark image

White image

Calibrated image

Active Pixel Sensor Dark current (physics)
124379
de