Image processing on mobiles can be pretty performance intensive given the relatively low power of their CPU’s (as compared to desktops).
In order to get reasonable processing speeds for my application Bokeh Booth I decided to build some efficient UIImage extensions which make use of the vDSP routines in the iOS 4 and up Accelerate framework. This framework allows for efficient matrix based image processing on iPhone/iPad devices, even those without OpenGL shader support.
You can grab a copy of all my routines here: https://github.com/gdawg/uiimage-dsp
The github repository contains a UIImage Category with the dsp processing routines and an example app showing how to call the routines.
I’ve tried to make the routines as easy as possible to use, for example you can blur an image using code as simple as
UIImage* src = [UIImage imageNamed:@"image.png"];
UIImage* blurred = [src imageByApplyingGaussianBlur3x3];
At this stage the library is still very much in it’s infancy but there’s support for blur/sharpen/2pass Gaussian Blur/emboss functions with the potential for many more to come.
The two pass Gaussian filter is particularly fast compared to any trivial implementation in software.
The github repository contains an example project showing usage:
If you use the library anywhere I’d love to hear about it. Also if you have any bugfixes or ideas on improvements I’d love to hear them too!
UPDATE (1/Mar/12): you may also want to check out CoreImage (if you’re only supporting iOS5+) and https://github.com/BradLarson/GPUImage for other options. I’ve not yet used either but the capabilities of GPUImage look particularly impressive and CoreImage looks amazing (for iOS5+ projects).