Does Compression Affect The Value Of Images as Evidence?

Click Here to Close this Window

    We see pictures pasted all over the internet in various formats, most are lossy. But just how much can this affect what we see? Let's take a look at the most common format, the "JPG" Image.


    What is illustrated here is a simplified version of what can become a very complex issue. You can clearly see how colors are altered and pixels changed when a digital picture is processed. To illustrate this take a look at Figure 1 below. It shows a three colored image, Dark Green, Red and Black. (Disregard the yellow box for now; it simply shows the central portion we will be explaining shortly.) There are only three colors present. The pixels in those colors are weighted as follows, Green="008000", Red="FF0000", and Black="000000". Every pixel in this figure is one of those three values.

    Next, the center section in the yellow box was copied and converted to a compressed JPG format. This was enlarged and shown as Figure 2. Note how the pixels near any edge has many color intensities present. These colors are false, they do not exist in the original image. They are artifacts created by the conversion process. The edges of the black line also have a fuzzy grey boundry, not the clear solid black of the original. Some of these false colors are close to the original and difficult to see, but others are clearly visible.

    Now for a better look at how significant this is, Figure 3 below shows a small section of Figure 2 along the line between the Red / Green boundary. You can see the multiple intensities of pixels that have been created. Finally in Figure 4 we have taken Figure 1, enlarged it for clarity, and changed only those pixels which are unaltered to Blue. What remains demonstrates just how significant the change was to the original image and how much damage was done to its value as evidence. Any area not blue is false, altered from the original.

    I should also emphasize this is a simplified example. Here, only three distinct colors were present in the original. In real life a picture will contain hundreds if not thousands of individual colors that comprise the image. Each of these would have an area around them similar to the ones shown in the example. Thus the number of possible errors will also be much greater as there are many more conversions done to compress the image. Likely the entire image would have false colors over large parts of it.

    A final note for any who wish to play with the examples presented here. If you view Figure 1 and enlarge it you may be inclined to question my claim of only three colors present, and you would be right. The posting of the figures on a website also creates errors. The original of the figures was processed using the non-lossy Bit Mapped Format. But websites require the use of jpg or similar formats which as explained, are lossy. Thus there are errors present even here. But the examples shown are still valid, they show errors actually captured using Bit Mapping and screen capture methods, which preserve the errors, not simply enlarging the image as you do on your computer.


Click here to Close This Window and Return to where you left off reading.   

© FEB 2014 - J. Brown . . . . . . .