One of the most popular features of security cameras is the ability to record only when something is happening in front of the camera. This has been the go to method for preserving that precious Hard Drive space, especially with resolutions always on the rise and with it recording storage demands. With event recording, you don’t have hours and hours of footage taking up space where nothing has happened. Up until now the most popular way of detecting an event is through motion.
How Does Motion Detection Work?
Motion detection is achieved by continuously monitoring each frame of video and averaging the background state of the scene, then if a significant portion of that scene (or percentage of pixels) changes from one frame of the video to the next, a motion event is triggered. For example, the camera is monitoring an empty hallway. Once a person enters the frame of the camera, the pixels that were of the floor and wall are now changed to the person. The camera logs a motion event and the recorder is told to begin recording. This technology was first introduced in analog security camera systems in the early-mid 2000s, but was fairly bad at giving accurate results. Unfortunately in the last decade, not much has changed.
Why Does Motion Detection Suck?
Motion detection events are still plagued with the same problems they had when they were first introduced. False positives from trees blowing in the wind, poor sensitivity, too much sensitivity, extreme shifts of lighting (i.e. a cloud moving in front of the sun), etc. have always caused this form of event detection to be less than useful. Some manufacturers have implemented the ability to exclude areas of the scene from triggering motion detection. For example, you can block out those trees that are always blowing in the wind, or the road traffic that is just visible at the edge of the parking lot. These features have helped, but never seem to work perfectly.
Another issue that needed to be solved was pre and post recording. In the early days, a motion event would occur and start recording, but you would inevitable miss the first couple seconds as the event was processed and the recorders Hard Drive spun up. Security Equipment Manufacturers solved this by building in a continuous buffer that always keeps the last few seconds or even minutes of video in memory. If a motion event occurs, the recorder knows to keep the buffered video and save it to the Hard Drive along with the footage from the point of the even on. But, even with these improvements, motion based recording still fails to be accurate enough to be useful in demanding surveillance and security applications.
So What Should I Use To Trigger Recording?
The advent of smart cameras and machine learning has opened up the market to a wide range of intelligent smart events that can be used to log and trigger many different aspects of your system. On the simple side of the spectrum, line crossing and region intrusion detection can be used in place of the relatively “dumb” motion detection. With line crossing, individual directions can be set to trigger and intrusion detection can require a certain amount of the region to be filled or that the object be present for a specified time threshold before an event is triggered. Some of the most advanced analytics include face detection, License Plate Reading and Logging, and People Counting. These advanced analytics can be used to more accurately monitor and interpret the actual events that are occurring in the surveyed area. Recording can now be triggered based on extremely specific criteria, such as if a specific license plate is detected, if there is a face seen, or if an object of significant enough size enters a specific region.
While motion detection has it’s place, it by no means is a complete or in many cases even a viable solution.
Contact an Arcdyn Sales rep today and ask about our line of smart cameras.