Deconvolution of a Black Hole

On April 10, 2019, the Event Horizon Telescope collaboration released the first image of a blackhole (designated M87), or more accurately the first X-ray image of the Black Hole's event horizon.  My first reaction was that this is really amazing.  The second reaction was, it's blurry.  The third reaction was to question whether SeDDaRA (or CARon) can improve the resoltuion to see more structure.  

To answer this question, the first step is to weed through the news sights in attempt to get a high resolution image that's as close to the actual data as possible.  Ideally you try to avoid images that have been compressed (like *.jpgs), trunacted digitally to fit into 8 bits per data point, and false colored.  To date, I (Jim Caron) have only found a "high resolution" false color image to operate on.  ( I will update this page if I can get something closer to the raw data.)  Processing this type of image can have pitfalls, but can also be enlightening.  For example, I have processed NASA images of their site where they had a smal image (like 32 by 32 pixels), up-sampled the image to something like 400X400, and then blurred over it to make it look better on news sites. 

The pitfalls are that you may reveal artifacts in the image that stem from the processing and not from the image data.  I produced the first (cropped) image below by converting the 4000X2330 color image (produced from a false color image) and converted to grayscale.  The second image was the result of applying CARon using a RofI=200, and a C2=0.01 to the first image.  The result is striking as much more detail is revealed.  This, however, is where one needs to be careful.  My guess is that this resulted from low bit depth of the original data, perhaps less than 8 (or it got translated into 8 bits), meaning there are fewer than 256 integer values. The group or their media team blurred the image and, applied the false color, and coverted it into an RGB jpeg image.  This result can, if one wished, to be regarded as a non-quantitative topographical image of a black hole.

Gray scale image converted from false color image of M87

This is a cropped 1091X937 portion of a gray scale image that was converted from a false color news release image of the Black Hole M87 event horizon.

 

This is the result of application of CARon, having the same contrast as the above image. While the result is striking, the structure could be the result of modifications to the original data.

All the steps above were performed using Tria's functions. The group used the alogithms CLEAN and Maximum Entropy for deconvolution of the data.  These are competing technologies for SeDDaRA.