We’ve all seen fingerprints and irises used as ID "badges", in movies if nothing else, but did you know that ears might work better? Evidence is mounting that ears might be the most reliable biometric feature to identify a person.
umut3502 was kind enough to share a conference paper he co-authored, posting a link at our forums. It features an ear-recognition system based on a mikromedia for STM32 M4, programmed with mikroC for ARM. All four authors are from the Istanbul Technical University.
Here’s the abstract:
In this paper, an embedded ear recognition system is proposed. System is based on STM32F4 Cortex-M4 microcontroller which is independent from a PC or an operating system. System is programmed with MicroC programming language which is similar to C or C++. DCVA has applied to the system for classification but before that for dimesionality reduction PCA has been applied too. During the realization of the system, many numerical methods have been used. System can store the dataset in its memory and this is the main advantage of it. This shows such an embedded recognition systems can be available with very small and cheaper products, as well. According to the experimental results, it does not take long processing time to finish all the processes.
Study the paper then think about all the equations next morning while exploring the unique shape of your ears with an end of a Q-tip. For any questions, try contacting umut3502 on the forums.