What "cycles" are you referring to? cpu cycles of the processor in the cam? I hadn't heard until now that one IVS function (intrusion zone) consumes more microprocessor cycles than another IVS function (trip wire)
How does one check how many cycles a feature consumes?
Maybe the burden on the processor is related to more than tripwire vs intrusion? For intrusion, are you using "Action" of "Cross" or "Appear"?
The camera *should* process the motion/tripwire/intrusion detection/<whatever> request, even if your camera is plugged into a PoE port of the NVR and you are setting up the camera through the NVR's software. That said, how NVR's and cameras work together is the subject of many a thread and a lot of frustration. Even if you have the same brand of NVR and camera, the constant flow of new devices, firmware upgrades and features means that whatever documentation you have for your NVR may work differently with one camera vs. another. It's a clusterf**k of combinations and permutations. If the NVR and camera are a different brand, good luck. In that case, if you can get, say, motion, it *may* be that the NVR processes it. It's just so hard to say. But it should be the camera.
Surveillance device companies (Hik, Dahua) are notorious for poor documentation on how you are supposed to achieve something, instead just documenting what some buttons in the interface do. No sooner than the manual is produced, the firmware changes. And the right hand doesn't talk to the left. One software development team for the camera software, another for the NVR. It's incredibly frustrating and a pet peeve of mine. But hey, you (sort of) get what you pay for.
On the actual cpu usage for different functions, it makes sense that intrusion detection is going to chew more cycles than motion detection, which may or may not chew more cycles than line crossing. With line crossing, the cpu will be looking at pixels on the line for a certain percentage of change (remembering that even with an indoor camera in a well lit, enclosed area like a doorway, the pixels will change color ever so slightly, but constantly, and there can be a lot in a hi res image). But it also has to look at the pixels around the line to determine which direction the crossing comes in. Bigger line? More work. For motion detection, only the specified area has to be considered, but it might be more pixels because an area is a lot bigger than a line. For intrusion, there's more work again. A certain percentage of the area has to change and maybe there are thresholds to cross.
So yeah, a really big intrusion detection area is going to take more to process than a small line crossing. But there are all sorts of algorithms out there that have optimized the crap out of the calculations so that in terms of complexity it's logarithmic O(log N), i.e. you make the area 10 times bigger but it only takes a little more time. Depends on the geeks that wrote it.
