Alien Anomalies

Members Login
Username 
 
Password 
    Remember Me  
Post Info TOPIC: Theory and practce of sharpening


Veteran Member

Status: Offline
Posts: 432
Date:
RE: Theory and practce of sharpening
Permalink  
 


humanoid, thanks for that tip!  I've used imagej, but did not know there was a pds plug-in!

qmantoo - that focus plug-in sounds interesting.



__________________


Teaching the truth

Status: Offline
Posts: 1921
Date:
Permalink  
 

There are also plugins for all types of manipulations that you may want to do as well as stacking and increased depth of field & focus plugins too.

Many of these plugins are developed by medical establishments for processing microscope images and medical imaging instrument results so are fairly stable and reliable.

However, the usual arguments apply as to whether we should use any of these different modifications to the nasa images to highlight different things.

__________________


 



Veteran Member

Status: Offline
Posts: 335
Date:
Permalink  
 

marsrocks wrote:
Using pds, I downloaded the img file; opened it in "Nasaview" software - then saved it back as a gif.


Another good program for opening uncompressed PDS image files that is free, cross platform and has a zoom function (unlike nasaview) is ImageJ. You will need to download and install PDS Reader plugin to make it work though smile


__________________


Veteran Member

Status: Offline
Posts: 432
Date:
Permalink  
 

I think this is worthwhile to compare pds vs. jpl site images, so here's mine.

Take a look at this comparison of an image sourced from the public jpl site and the pds site:



This is a rock we spent some time with in a previous post.  Look at how different it appears when we can get rid of so much of that black unviewable area:




JPL has already overstretched the image - while pds has a more even distribution of pixels across the grey scale.

JPL original is here:

http://marsrovers.jpl.nasa.gov/gallery/all/2/p/514/2P171996132EFFAALCP2427L7M1.JPG


PDS for spirit, by sol at this link:

http://pds-geosciences.wustl.edu/mer/mer2-m-pancam-2-edr-sci-v1/mer2pc_0xxx/data/


Spirit Sol 514 at this link:


http://pds-geosciences.wustl.edu/mer/mer2-m-pancam-2-edr-sci-v1/mer2pc_0xxx/data/sol0514/

PDS selected *.img image is downloadable at this link:

http://pds-geosciences.wustl.edu/mer/mer2-m-pancam-2-edr-sci-v1/mer2pc_0xxx/data/sol0514/2p171996132effaalcp2427l7c1.img



Using pds, I downloaded the img file; opened it in "Nasaview" software - then saved it back as a gif.






__________________


Dedicated to the truth

Status: Offline
Posts: 1099
Date:
Permalink  
 

marsrocks,

I already have shown 'before and after' examples in this thread of an image captured on Mars.

I'm waiting for O'Brien to post an example of what he can achieve to improve a similar image.

__________________

"All truth passes through three stages. First, it is ridiculed; Second, it is violently opposed; Third, it is accepted as being self-evident."

Arthur Schopenhauer



Veteran Member

Status: Offline
Posts: 432
Date:
Permalink  
 


Something like this:

before and after sharpening (from OBrien's article):

beforeandaftersharpenin.jpg

__________________


Veteran Member

Status: Offline
Posts: 432
Date:
Permalink  
 

Why don't both you guys put up labeled side-by-side comparison images to prove your points.

Thanks.





__________________


Dedicated to the truth

Status: Offline
Posts: 1099
Date:
Permalink  
 

The first thing I would like to say is that this forum is not a strict science forum where everything has to be proved mathematically. There are many readers who visit forums, such as this one, who are not interested in the quality of the imaging. On the other hand, there are other readers who are interested in viewing good quality material.


Timewarp wrote:

"Then, by applying a controlled amount of sharpening the overall appearance of the image can be improved."

In reply, O'Brien wrote:

"Appearance improved by altering the content."

With respect O'Brien, you are incorrect. The original content is still contained within the image.


O'Brien wrote:

"Then quit using jpgs, and quit mislabeling them as "originals" when thay are public web conveniences."

There is nothing wrong with using the jpeg images as I personally know of many universities who use them for analysis. The downloaded images are not mislabelled as they are the original jpeg's made available for downloading.

 

Some of the PDS material available in many cases is not the ideal for sourcing a reasonably clear image. In the past I have examined a high definition PDS tiff and a high resolution jpeg and found the jpeg far superior to work with than the large tiff file.

I understand perfectly what O'Brien is saying, but there is no requirement for using PDS material as the source material to post images on the forum. The name of the game is to produce an image of reasonable quality in order to identify some of the anomalies and other surface features. In these particular circumstances using PDS material with huge file sizes is not the best way to achieve that goal.

In the images I posted above it can be seen by applying the 'gaussian blur' function to a downloaded image that the clarity is improved straight away by removing the disturbing compression artifacts. This procedure followed by a slight amount of sharpening to enhance the edge definition in no way erases or adds extra content to the image.

I would be interested to see if anyone can produce the same cropped section clearer than I have shown with a full description of the procedure implemented to produce their image.

With reference to the material posted above, would members say the sharpened images are an improvement on the 'raw' image? Also, is any of the content missing in the improved images?


Here is the reference for the image I used in the example.

http://marsrover.nasa.gov/gallery/all/1/p/2379/1P339388999EFFAS00P2418L2M8.JPG

 



__________________

"All truth passes through three stages. First, it is ridiculed; Second, it is violently opposed; Third, it is accepted as being self-evident."

Arthur Schopenhauer



Member

Status: Offline
Posts: 184
Date:
Permalink  
 

Timewarp wrote:
The examples shown below give an indication of how the jpeg compression artifacts can be removed ...

Smoothed.

Timewarp wrote:
... Then, by applying a controlled amount of sharpening the overall appearance of the image can be improved ...

Appearance improved by altering the content.

Timewarp wrote:
... to reveal the content ...
Low frequency is retained. High frequency content is altered.

 



Timewarp wrote:
The softening/sharpening process is required as the quality of the downloaded jpeg originals is inferior for extended research and analysis.

Then quit using jpgs, and quit mislabeling them as "originals" when thay are public web conveniences.

The inability or unwillingness to use the best available images from the PDS, and to steadfastly maintain they are "lacking in quality that's why I prefer to use hi-res jpeg's if available" and then to try to insist that they have no change in content shows a deep lack of understanding of image processing techniques and theory.

There seems to be a pervasive misconception that a smooth image is a better image, and so anything that can be done to a jpg to smooth out compression artifacts somehow restores the information lost int he compression process. Smooth images are better in print advertising and photography magazines. Not in planetary science.

It seems as if I do have to repeat myself incessantly. DO use PDS images if you're trying to work at a resolution near the pixel level. DO use PDS images if you want the correct brightness and contrast levels. DO use PDS images to get metadata about the image. DO NOT use jpg images other than for convenient references or pretty pictures. DO NOT zoom into a jpg image 500x and rely on the outcome as accurate.

Image compression loses information. Blurring loses information. Sharpening loses information.

 


 


 



__________________


Dedicated to the truth

Status: Offline
Posts: 1099
Date:
Permalink  
 

The examples shown below give an indication of how the jpeg compression artifacts can be removed from the majority of downloaded images. Then, by applying a controlled amount of sharpening the overall appearance of the image can be improved to reveal the content without a blanket covering of compression artifacts.

The softening/sharpening process is required as the quality of the downloaded jpeg originals is inferior for extended research and analysis. No other processes have been applied.

Downloaded full-size image showing area to be examined.

test_1P339388999_section.jpg


Crop showing compression artifacts.

test_1P339388999_raw.jpg


Removal of compression artifacts using 'gaussian blur'.

test_1P339388999_gb.jpg


Sharpening the image (2 independent 'sharpens').

test_1P339388999_gb_2s.jpg


Sharpening of image (3 independent 'sharpens').

test_1P339388999_gb_3s.jpg



__________________

"All truth passes through three stages. First, it is ridiculed; Second, it is violently opposed; Third, it is accepted as being self-evident."

Arthur Schopenhauer



Teaching the truth

Status: Offline
Posts: 1921
Date:
Permalink  
 

There appear to be at least two issuues here. One is that we want to be like a forensic scientist and find anything that is possible from the 'scene' and the other is to bring out as much detail as we can without altering the fabric of the original, which would invalidate the evidence for the 'courts'.

Then a number three, there is the fact that a small sample is often tested to destruction to yield extra facts about the whole thing.

Looking at it like this would support both views.

The issue number 3 would allow for the softening and sharpening technique and anything else which may give some insights into the general image but cannot be used for "evidence in court". Having theories about life on Mars is useful because we can look for evidence in other photographs, and see the relationship of one feature to another. For example, I have learned to look for these doughnut-like shapes as evidence that there is a community or presence around that area. It seems to be a flag or signal to indicate this, similar to a shop sign or house name.

Issues number 1 & 2 mean that we strictly cannot alter the image in any way, however, even altering the brightness will alter the values of the pixels at the lowest level, so it boils down to the degree of alteration. The images have been altered from their raw state anyway and have been 'processed'  before we get to see them, so how much does our processing effect the image? It is all a matter of degree, isn't it?

Technically as I understand it, at the pixel level, the sharpening algorithm looks at a square group of pixels and, where it detects an "edge" in the block of pixels, it will accentuate that edge. The softening algorithm will look for "edges" and make the difference between the pixels less, so that the "edge" becomes less defined.

All this definitely has some effect on the values of the pixels so your image will change. Sharpening will produce the effect of things that were not so defined becomming more visible. The increased contrast either side of the "edge" found will be more pronounced and softening will do the opposite, but not necessarily the exact opposite. Software differs in the amounts of sharpening and softening and in the algorithms they use for each.

So whereas the overall effect will appear as if the image compression artifacts are removed, what it has done is to soften the edges of those artifacts and other image "edges" (like rocks etc) by an amount and then sharpening will make these, and other edges, more defined again but not necessarily by the same amount.

Both processes do alter the overall image and will lose some and introduce some edges (not only compression artifact edges) that were possibly not there before you did this process. Think of sharpening as an accentuation of contrast between adjacent pixels in some cases.

Unfortunately, making "something" out of "nothing" is better posted on an art forum as it does not increase the evidence for life outside Earth.



__________________


 



Veteran Member

Status: Offline
Posts: 400
Date:
Permalink  
 

Obroen
-----------------------
Since "image sharpening" is so
often used (and misunderstood)
in the study of planetary
mission images, here's a little tutorial for novices in easy, lay terms. The take-home message
from the article is "It's
important to realise that
sharpening degrades your
image by introducing distortions
that weren't in the original scene".
----------------------

Agreed,,

__________________


Member

Status: Offline
Posts: 184
Date:
Permalink  
 

Timewarp wrote in "Aerial views of ivilization evidence." [bolding mine]:

I have used the PDS material and found it lacking in quality that's why I prefer to use hi-res jpeg's if available, although it has to be said that the PDS reference material may be useful for making comparisons.

The images posted above are the jpeg's but contain no compression artifacts as they have been removed. What can be observed is the RAW content.

The program used to process and improve the images is Paintshop Pro v 6.0. When using this application to remove artifacts first load the original image, select 'gaussian blur' then use 'sharpen more' followed by 'sharpen'. This basic amount of filtering is sufficient to view the majority of jpeg's without any change to the image content.


Claiming no change to the image content in this case is utterly and absolutely false. The curious reader should review any text on information theory to gain knowledge of some of the basics of this topic.

Since "image sharpening" is so often used (and misunderstood) in the study of planetary mission images, here's a little tutorial for novices in easy, lay terms. The take-home message from the article is "It's important to realise that sharpening degrades your image by introducing distortions that weren't in the original scene".

 



__________________
Page 1 of 1  sorted by
 
Quick Reply

Please log in to post quick replies.

Tweet this page Post to Digg Post to Del.icio.us


Create your own FREE Forum
Report Abuse
Powered by ActiveBoard