Iris Finder
The Iris Finder program is designed to process photos of human eyes taken in the visible light.
The tool provides a reliable algorithm for localizing inner and outer iris boundary in the jpeg image.
It is written in plain class-based manner in C++, utilizing IPL library for image data structure and FLTK
library for user interface creation.
The piece of software features a graphical user interface for loading the images (or whole directories) and visualizing the effect of localization. Moreover the applied algorithm is optimized for speed in a way it is capable of tracking the iris in real-time, while a series of images is streamed from the camera. The main steps of the algorithm are shown in the diagram below.
The piece of software features a graphical user interface for loading the images (or whole directories) and visualizing the effect of localization. Moreover the applied algorithm is optimized for speed in a way it is capable of tracking the iris in real-time, while a series of images is streamed from the camera. The main steps of the algorithm are shown in the diagram below.
Main Window with the processed image loaded.
Recent Changelog
Version 1.2
Version 1.2
- Localization of eyelids
- Real-time iris tracking > Watch movie
- Controller for the camera movement via IrisComm library > Watch movie
- Autofocusing of the camera
1. Read in colour image from file.
2. Find reflections.
3. Fill in reflections in colour image based on RGB values of neighbour pixels.
4. Convert colour image with filled-in reflections to grayscale image.
5. Reduce red component image size and roughly find inner iris boundary based on Daugman's integrodifferential operator.
6. Find outer iris boundary precisely in grayscale image with original size based on Daugman's integrodifferential operator.
7. Find inner iris boundary precisely in red component image with original size based on Daugman's integrodifferential operator.
After the iris is localized it can be transformed from the carthesian coordinates to the polar coordinates.
Iris projection.
Reflections map: black area - pixels with actual iris image, white area - reflections, to be removed for extracting iris features.
