CODEWITHSUNDEEP
c#image-processingemgucv
Andrew Simpson
Andrew Simpson

Reputation: 7314

removing the minor variations in pixel intensities in an image

using C# and emgu.

I am working with jpegs which will ultimately end up in a browser.

I am already lowering the quality of the image (to 60%) to reduce the byte array size.

Whenever I get the chance I spend sometime looking around to find ideas to reduce the size of this byte array even more.

I do know the brighter the image the more bytes it seems to hold and the more contrast the image has the more bytes it seems to hold.

Upon googling I came across this:

http://en.kioskea.net/download/download-666-greycstoration

It seemed to imply that by reducing minor pixel variations in an image that I can reduce the byte array defining these jpeg images.

So, to my approach (and understanding)...

Do I iterate through all the pixels and 'average' a group of pixels say on a 4x4 area? Or am I missing the meaning entirely here. I ask because I have already done this but it makes no difference to the image size (in bytes).

I could post my code (and will) but it was just a mock up and not fit for production code.

I am more interested in understanding the meaning/implementation of all this. I can code it this myself as soon as this is clarified (and will post back with the code here)...

Upvotes: 0

Views: 150

Answers (1)

NDEthos
NDEthos

Reputation: 516

Could not add comment due to rep points.

First, if you are going to do and average or some type of filtering on a per pixel basis while using nearest neighbors then you will most likely want to use a filter that is an odd number say 5x5. This will cause the target pixel to be centered in the filter.

You will end up with smaller or equal sized files but only testing will prove how much smaller.

EDIT:

So I think you could also use a simple reduction factor that would give you less colors per image. The algo that I saw was data[i]= (data[i]/N) * N + N/2, N is the reduction factor and data[i] is to be each component of each pixel in your image. The pixels will have the same range 0-255 for each of the rgb channels but certain numbers in this range will not be available to the pixels depending on the reduction factor!

This should give you a reduced jpeg image because you will have less colors.

Upvotes: 1

Related Questions