The place I work at recently installed a biometric time clock. You punch in your 3-digit ID number, then press Enter. The machine then asks you to place your hand on a little platform studded with a few metal pegs. When your fingers touch the pegs the machine beeps and clocks you in or out. At the same time the display shows a score. I’ve had scores as low as 1, and as high as numbers in the 30s.
My coworkers and I have a running debate as to what exactly the score means. We all pretty much agree among ourselves that it indicates how good a match our “handprint” is to the original reference print we gave when they first installed the clock. (We have no solid proof for this theory, but no one has come up with anything better.) We cannot agree, however, if a low score or a *high *score means the print is a good match. Anyone know?
Before anyone asks, No, I do not have a brand or model number for the clock. I’m hoping the score readout is a universal feature of all time clocks of this kind. If not, please set me straight.
My company expermented with that type of time clock. It was explained to me that the score was an indication of how well your hand print scanned. I have no idea if that is acuate or not, I got it as second hand information.
My company did not continue to use these very long. We went back to hand writted time sheets.
The company I work for uses these, only it scans one finger rather than the entire hand. It scores between 1-100, comparing to the original fingerprint taken when it was installed, and the higher, the better. (I currently hold the record, at 96 :D).
See if you can google around for a manual. The system I beleive your talking about is called “Handpunch.” An interesting fact about it is that everytime you use it, it updates the image, this makes it possible to continue to recognize someone if they, for example, add or lose alot of weight.
Also interesting, a friend of mine recently had upgraded to a fingerprint system and then had to tear it out and install a handpunch system. He employs a lot of Mexican workers that refused to let him take their fingerprints (INS issues I assume), the handpunch system takes a picture from above, so no fingerprints on record.
This sounds like the same ADP handpunch that we have. I actually installed it and run it at my workplace. A lower score is a better “match” to the handprint template in the system. The default settings of the handpunch allow variances up to 130 (i believe), but you can adjust the settings to be LESS accurate if people are having problems punching in.
If I may: what’s the point of requiring you to punch in your employee code first? Shouldn’t whatever biometrics the system reads be sufficient to positively identify you? The system in use where I work scans fingerprints instead of handprints, but you only need to enter a badge number if the scanner can’t ready any of your prints.
Our handpunch system is sufficiently less sophisticated than our biometric fingerprint readers. My guess is that the handpunch only checks to see if your hand matches your template, while the fingerprint system scans all prints in the computer to see if it matches any of them.
I can’t tell you for sure - because each biometric algorithm is different depending on who developed it.
However, from working with fingerprint devices I can tell you that in general, the higher the score, the closer the match. In fact, I have never worked with a device where their algorithm behaved differently. Every device I have worked with has given a higher score for a closer match.
This speeds up the matching process. The fingerprint or handprint is stored as a binary buffer which cannot be indexed. This means that you have to look at each one to see if it is a match and worst-case scenario would have to look at every record to find a match. If you first enter an ID which *can *be indexed, you only have to retrieve *that *record to see if you have a match. In developing login systems there is always a big discussion on usability - with many of the domain experts making the argument you have and some arguing for speed.
I usually say that unless you have millions of records, you need not worry about speed. Even several thousand records can be searched relatively quickly, IMO.
Some will argue that this also makes it more secure. There is a very very small chance of a false-positive. (I forget the numbers but have a book some-damn-where that might tell me.)
If you have to input a secondary ID, you should eliminate the false-positive. I have worked with places that wanted *three *verification methods: An RFID tag, a password and a biometric ID. These were very secure government buildings.
Well, I repair and work with these things on a daily basis. The hand punch “score” is really as many of you have guessed is how well the current “snapshot” matches the stored template in the clock. Also, the template is updated every punch, therefore the clock takes an “average” among every punch you put in. the more punches, the better the template. Here is where the score comes in. The score is exactly what you think it is, it’s how well your current “snapshot” matches against that rolling average. the lower the score, the less deviation from the current template. As the clock gets dirty, the score will start to get higher, until it can no longer match the template. When this happens you’ll see scores upwards of 80, 90, and even over 100… A score of 1 is optimal, which means there is only 1 error between your current “snapshot” and the stored template.