Thursday, October 1, 2009

Finger Image in Unique ID Project - how the accuracy and search will be achieved

JPEG Image




BMP Image



Referring to the post in the following link http://www.hindu.com/2009/09/27/stories/2009092755850900.htm

That contains following lines:

“If any agency wants to confirm the identify of a person, it would have to just take the fingerprint of the person on a cell phone and send it across to a central database and receive authentication within seconds.”

UID project is one of the biggest IT projects of the world. There are many identified and unidentified risk areas. Image work associated with the Finger image detection which will identify the uniqueness of any individual in biometric identification needs deep attention.

Till date there is no image search engine around the whole world. Few of the dedicated research labs are working towards the same. We are very optimistically talking about capturing the image from an ordinary low-end handset, may be a high-end handset where the high-level mega pixel camera is used.

To my knowledge in every handset/camera the image generated is captured in the proprietary raw format and rendered into a common format e.g. JPG which is very widely used.

Of all the available images formats– JPG – the most common standard format is a COMPRESSED format. (Data is lost in this format) For biometric identification we need accuracy that means we need raw data format e.g. BMP format (Lossless data format) format or PNG format for storing the image in the central repository. Most of the scientific work involves TIFF image format.

SVG is a new image format, it is scalable but to my limited knowledge - how many handsets would convert the image in this format is the key.

Now if we take JPEG format, an average image takes say 5 KB of space. 60 million images mean 60 million*5 KB of database space in the central repository.

Besides security, identification, authorization, authentication, database design, backup, application failover mitigation, 24 *7 availability, adaptability – in case of power failure, natural disaster, adaptation, failover and security issues under low network bandwidth, flexibility to scale up or scale down the whole software architecture, finger image scanning in few seconds over the provided network bandwidth is the key to the success of this project.