A new JPL project requires that the center of the images of a rising or setting Sun be determined with certain accuracy — for example, within 1 km — when the Sun is observed from the International Space Station (ISS). This center-finding technique can be used in applications such as a Sun-Tracker. To meet such needs, a simulation tool was developed for the generation of Sun images observed either on the ground or from space. The new technique enables one to find the center of a Sun image based on simulated images. The technique does not rely on ellipse-fitting to the boundary of a Sun image or other calibration techniques, so the accuracy is not affected by the distortion of Sun images.
When observing the Sun image from the ISS, the Sun exhibits an equivalent diameter of approximately 21.5 km. So the 1-km accuracy of the Sun image center translates into a pixel-based accuracy. The longer axes of the asymmetric Sun images are not aligned with the horizontal or the vertical axis. Before finding the center of an image, the angle of rotation of an image is found such that the image’s longer axis aligns with the horizontal or x-axis.
This was accomplished in the following way. A tilted image is normalized such that its peak = 1. A threshold is applied to the image to generate a binary mask and find the approximate center of the image by calculating its center of mass. The binary mask is rotated, and a – b is calculated, where a and b are the larger and smaller widths of the mask, respectively. The angle of rotation is determined from the location of the maximum of a – b .
The widths of the simulated image with ray vertex height hv = 22.5 km have the same [Width Height] values as the measured image. Because the exact center of this simulated image is known, it can be assumed that the center of the measured image is located in the same location, measured from the left vertex to the center of the mask, as well as from the top vertex to the center of the mask.
This work was done by Erkin Sidick of Caltech for NASA’s Jet Propulsion Laboratory.