Corrections are made for bias and noise.

A method of processing the digitized output of a charge-coupled device (CCD) image detector has been devised to enable reduction of the error in computed centroid of the image of a point source of light. The method involves model-based estimation of, and correction for, the contributions of bias and noise to the image data. The method could be used to advantage in any of a variety of applications in which there are requirements for measuring precise locations of, and/or precisely aiming optical instruments toward, point light sources.

The principal sources of centroid error are bias and noise in the outputs from the pixels of the CCD. Noise consists mainly of fixed components (readout noise and noise from dark current) and variable components (pixel defects and shot noise from background light). Bias is caused mainly by stray light and nonuniform distribution of light in a background image. Image

In the present method, prior to normal operations of the CCD, one measures the point-spread function (PSF) of the telescope or other optical system used to project images on the CCD. The PSF is used to construct a database of spot models representing the nominal CCD pixel outputs for a point light source projected onto the CCD at various positions incremented by small fractions of a pixel (see figure).

During normal operation of the CCD, the centroid of the image of a point source of light is initially computed from the digitized CCD pixel outputs in the conventional way. However, this initial computation of the centroid is used to retrieve the corresponding spot model that was constructed earlier. Then the boundary between noise and signal is determined by comparing the spot model with the CCD pixel outputs. Pixel positions of same pixel value of the spot model and the image data in the background area are defined as the boundary. All pixels of the image data beyond this boundary are set to zero. This effectively removes the noise and bias in the subsequent centroid estimation from the corrected image data.

This work was done by Shinhak Lee of Caltech for NASA’s Jet Propulsion Laboratory. For further information, access the Technical Support Package (TSP) free online at under the Information Sciences category. NPO-30585

This Brief includes a Technical Support Package (TSP).

Reducing Centroid Error Through Model-Based Noise Reduction (reference NPO-30585) is currently available for download from the TSP library.

Please Login at the top of the page to download.

The U.S. Government does not endorse any commercial product, process, or activity identified on this web site.