Biometrics Using 3D Vision Techniques

Scientists claim that sometimes just 20 seconds are enough time in which to recognize a person. However, the human ability to recognize a person is quite limited to people who are closer or that have been known or associated with. On the contrary, an automatic identification system may require a longer time for recognition, but is able to recognize people in a larger set of potential identities. Nowadays, the most popular biometric identification system is certainly represented by fingerprints. However, present biometric techniques are numerous and include, for example, those exploiting hand shapes, iris scanning, face and ear feature extraction, handwriting recognition, and voice recognition. Early biometric systems were implemented by exploiting one-dimensional (1D) features such as voice sound, as well as two-dimensional (2D) ones such as images from fingerprints or the face. With technology advances and cost decreases, better and better devices and capture techniques have become available. These have been investigated and adopted in biometric settings too. Image vision techniques have not been excluded from this evolution-supported acceleration. The increase of computational resources has supported the development of stereo vision and multiview reconstruction. Later, three-dimensional (3D) capture devices, such as laser or structured-light scanners as well as ultrasound-based systems, were introduced. The third dimension has therefore represented a significant advancement for the implementation of more and more accurate and efficient recognition systems. However, it also presents new challenges and new limits not yet overcome, which are the main motivation for the research in this context.