How I Did My Image Processing
Simple steps that will allow you to do it yourself - part 3
Now that we have some decent image data to work with, we can try a few other tricks to sharpen or enhance the images. First, we are up against the resolution limits of the microscopic imager. Is there a way to do better? It would sure be great is we had more pixels to work with in our image data.
There is a simple method to do this- it depends on having more than one picture to start with. This method is called frame stacking. Imagine that you have a camera and you take four or five pictures of the same scene. Each image will be slightly different, even though they are taken from the same viewpoint and at the same resolution.
Random noise can occur in any imaging process, but if you take a number of samples, you will find that the chance of that noise occurring in any given pixel is small. By adding a number of those identical pixels together, you can "vote out" the noise signals and strengthen the actual image data. This is the first advantage of frame stacking.
| We will
need multiple pictures of the original image to start the process
out. I see that there are five nearly identical pictures of this
object, so we have plenty of data to work with.
The first step is to take the largest image and scale it to 200%. This allows us to eliminate some of the JPEG errors and to also make room for averaging in the new frame data. We will also need to measure each "Jupiter" fossil and scale them to all be exactly the same size as the newly enlarged image we start with. After each has been edge enhanced (as before) and then scaled to the same size, they are overlaid and stitched. The fifth picture was left out because it was slightly out of focus. I used the first four images only. Even with this magnification, you can now see much more detail and the pixel edges and JPEG artifacts have been erased. We have effectively just about doubled the resolution of the MER-B microscopic imager and gotten a better image than NASA has in their raw data. |
 |
|
If I now use other mathematical methods to find details that are present but invisible to our eyes, we can see subtle variations in shadow and shape that would be impossible to locate in any other way. One such method is a cosine transformation. By converting the individual pixels into numeric values and then performing a cosine operation on each, I can extract faint changes from one point of the image to the next. After this transformation, the numbers are all converted back to pixel values and become a new image. The resulting image looks strange, but it is a display of the variations in the image data based on the cosine transformation. Looking at it, we can see some odd things that were not very clear in the old image- but if we look closely, they are definitely present in it. Now we have to add it into the old image and see how it enhances the visibility of the features. |
 |
|
| And here we have
a very clear view of the finer structures of this spherule. There is
something that looks very much like a bulls-eye on the right of center,
like a rounded dot with radial lines coming from it. It also has a
faint circle running around it.
What is it? It is impossible to say, but it looks less like a mineral as we begin extracting these hidden features. Also, there are features that this process has obscured, but patience and further work can reveal those as well. By inverting the cosine data and overlaying it on the original picture once more, we can get a better view of the underside of the organism. |
 |
|
| This image is
washed out at the top, but the bottom now reveals a rounded feature that
was difficult to see in the original. Every "spherule"
shows this or similar features, and they are identical to the mouth or the
anus of the sea urchin.
In other words, these "rocks" have mouths and anuses. Why would a rock need that? In my experience, if a rock has an anus, it is either a statue or a fossil. |
 |
It is easy to see that other processes or refinements of these can be applied to obtain excellent images, even in the face of fairly noisy or poor data to start with. These methods are used by scientists, law enforcement, and industrial spies every day. They are proven to work, and they have yielded some of the best images of the Martian fossils seen.
Using this and similar methods, we can find out far more than a cursory glance will show us. Almost any photo editing program can do these operations- Adobe Photoshop can as well as cheap scanner software such as IP Plus. Try some of these methods and you can verify what I have found for yourself, and you might also find something that I have missed.