Wavelets are used for pattern recognition in many iris recognition systems. First, the position of the iris in the input image is determined, then any eyelash and eyelid occlusions are removed and the iris is extracted by converting it to a polar coordinated image. The resulting image signal is convolved with wavelets of different shapes and sizes. The resulting signal is encoded using phase demodulation to produce an IrisCode. Two iris codes can be checked for a match by computing their hamming distance. [1] is the paper which describes the original system invented by Daugman. [2] is an open source implementation of that method.
That paper by Daugman was the basis for my undergrad thesis, so when I saw the title, that's what my mind immediately jumped to! Lo and behold, I open the comments and find a link to the paper. Not only that, but it's still stored on my old professor's website (~az = Andrew Zisserman)!
[1] https://www.robots.ox.ac.uk/~az/lectures/est/iris.pdf [2] http://iris.giannaros.org/