Ceres has been the most educational object to process thus far. The input images are the same as the Jupiter detection input images, but the processing was way more critical for this detection than it was for Jupiter. Ceres really helped extend my toolset and ability to detect fainter sources in general. I had to develop some neat new tools like brightness matching to handle more subtle difference image signals that resulted from input images with significantly different relative brightness levels.

This detection also comes after the LSST Project & Community Workshop which I attended in August (as maybe the only unaffiliated amateur?). I learned a lot of terminology in the Solar System object detection domain along with some very helpful concepts I can use to further this project. For example, I learned that the LSST pipeline uses a difference then threshold method for difference image creation whereas I had been using a threshold then difference method. The LSST technique makes more sense as it results in a better behaved distribution of values in the resulting difference image and is what I am doing in this analysis and going forward. This was just one of many practical takeaways from the meeting. Anyway, below is the documentation of my Ceres detection from an AllSky camera.

1. The starting point for finding Ceres is to crop out Jupiter (the large moving blob) in the original two observations. Ceres is small and faint in these images and Jupiter's size and brightness overwhelms the difference thresholding. These images are the aligned image set 1 images from my previous Jupiter solution with a PSF deconvolution applied. The area investigated in the rest of this process is the crop within the red box.

2. Here's the red box crop from image set 1 with brightness matching applied to the dimmer frame. Notice the difference in brightness between the two frames in image set 1 compared to the difference here. The static sky template is generated from these two frames. Click and hold the frames above to see the static sky template.

3. Static sky template differenced frames [image set 2 frames - static sky template] with a 5σ threshold applied. Object blobs with a contiguous pixel count of 2 or less are removed. Linkable moving objects are in red.

4. Linked object detection translated back onto the frames from image set 2. That's Ceres motion over 27 days.

5. I've tried to duplicate the Stellarium crosshair here on one of the Ceres frames. The 3 star belt northwest of Ceres (circled) is the most apparent feature in Stellarium in image set 6.

6. Stellarium position of Ceres on 5/28/2019. I rotated the image to represent the perspective in image set 5 thus the vertical labeling.

Process notes:

Most of the process enhancements below were implemented specifically to detect Ceres. I used almost all of these techniques for Jupiter too, but they weren't required in that case. I'll reiterate these new techniques even though most were mentioned previously for Jupiter.


Published: 9/29/2019