Iris recognition based on statistically bound spatial domain zero crossing and neural networks

The iris pattern is a crucial biological feature of the human body. Recognizing individuals based on their iris pattern is becoming more popular because of the unique pattern each person has. Iris recognition systems are getting a lot of attention because the detailed texture of the iris makes it very reliable for identifying people. However, there are still several challenges when it comes to recognizing irises in less controlled environments.

This article discusses a highly reliable method for creating a biometric recognition system based on the iris of the human eye. Current iris recognition algorithms struggle with localization errors and the significant time required for this process. While spatial domain zero crossing is the simplest and least complex localization method, it hasn’t been used much due to its high sensitivity to erroneous edges. Instead, more complex and time-consuming algorithms are typically used. By applying appropriate statistical limits to this method, it becomes the least error-prone and fastest option. Using this approach on the CASIA v1 & v2 datasets, errors were reduced to just 0.022%, and it required less time than other algorithms. Most algorithms require multiple comparisons to account for translation and rotation errors, making them time-consuming and resource-intensive.

This method uses a single comparison and achieves a recognition accuracy of over 99.78%, as demonstrated by tests.

The technique involves using a neural network trained to recognize unique statistical and regional features of each person’s iris. The algorithm also accounts for illumination errors, elliptical pupils, excess eyelashes, and poor contrast.

Read this article here;

For publishing scholarly article in Bentham journals, please visit: