Identification, please

Turn to biometrics when accuracy counts

Quick, accurate identification of people is important anywhere security is important. Biometrics measures unique characteristics, such as fingerprints, palm prints, facial features, or retinas, to identify a person and possibly control access to locations or equipment.

Horizon Imaging uses various methods to extract biometric features from images. The images are then classified by feature set and entered into a database representing the original images but in a data-reduced format. This can be done using a neural network.

Then test images can be compared with information in the database to establish identity.

In one palm-print application, for example, Horizon Imaging used wavelets (features), principal component analysis (dimensionality reduction) and neural networks (classifier). Further research and refinements were presented at an SPIE conference.

If you need to determine identification, possibly to control access, Horizon Imaging can provide your solution.  

Tools of the trade

A few of the tools we use for biometrics applications:

• Neural nets / SOMs
• Morphology
• Edge detection
• Feature extraction
• Pixel rotation
• Hough transforms
• Wavelets
• FFTs
• Radial basis functions
• Color models: RGB, YUV, HSI, LAB