Question about piezo element impact sensor.

I've done the basic knock sensor before, and I'm planning on implementing it in my project which triggers a camera when an impact is detected above a certain threshold. However when I grabbed the piezo element that I had laying around (ripped out of a $1 door chime), I was experiencing something unusual. I hooked the piezo up to my multimeter to verify polarity, and according to my multimeter it seems to be producing both positive and negative voltage interchangeably. Its hard to tell since the multimeter refresh rate isn't super fast, but It almost seems like if I flex the piezo, it produces a positive voltage, and when I let it unflex, it produces a negative one. If I simply give it a strong impact, sometimes it shows up positive, sometimes negative.

All my research has always indicated that piezos only have one polarity, positive in the center, negative on the outside. So why is this one not matching up, and is it safe to use with the arduino?

Also, are piezos safe to use with arduino for very strong impacts? Or could a very strong impact potentially produce too high of voltage? (I'd be using a 1MegaOhm resistor like in the tutorial) I want to not only detect somewhat sensitive impacts like knocks, but also very strong impacts such as having the sensor on the handle of a baseball bat or hammer that hits stuff, or strapped to the side of rifles or pistol slides to instantly detect the strong recoil.

PS: Is there any other cheap zero delay impact sensor I might want to consider that could be better, or is a piezo the best bet? Doesn't have to accurately measure the "strength" of the impact, only that the impact is present above a set (very low) threshold.

Piezo elements are polarized in that a given strain direction leads to a given electric field direction - if they didn't come with polarity marked you couldn't guarantee two units would be in-phase. The electric field (and the strain) can be in either direction.

So are you saying that straining or flexing the piezo in one direction produces positive current, and flexing it in the opposite direction produces negative current? Meaning a strong impact, which would cause it to vibrate, or distort back and forth very fast would be producing both positive and negative current? That seems to be what mine is doing, although that does not match up with everything I have read.

Meaning a strong impact, which would cause it to vibrate, or distort back and forth very fast would be producing both positive and negative current?

Yes that is correct.

although that does not match up with everything I have read.

The internet is full of rubbish. The 1M resistor allows the internal protection diodes some space to stop the signal zapping the input.