|
Smart Video and Object TrackingBy Rod Deluhery |
|
OP911.NET tracks down the latest in video systems. We look at the latest in object tracking, motion detection, image processing and video management.
As security cameras multiply, the ability for people to successfully monitor them declines. Commercial and public agencies are looking at monitoring hundreds of camera feeds with the minimum amount of staff. Many agencies are already surpassed their 1000th video camera install. How do agencies manage and store all that video? Let’s look at how some different techniques to manage video. Smart video systems are sold on the premise that one person can only do so much, and it’s cheaper to have a computer look at the video than the security guard. Simple, right? Well not really. Here we look at several techniques on the market that help you stretch your budget. You may not be able to replace that operator with a machine, but you may not need to hire that other operator. Ok, well that is not what you want to tell your staff, but essentially that is the goal. . .do more with less money.
Video motion detection. . . why do I need it? Digital motion detection is the processing of video images to detect differences between the video frames. Note this might not be actual motion, sometimes it could be that the camera was actually moved by wind or an intruder (like the intruder taking the camera!). But what normally happens is that objects in the scene move, and a system detects the changes. Then you can do two things. That video can be marked as “motion” video. The motion can also be setup to trigger an alarm, or other action. . .say for instance trigger the activation of other recording devices.
There are other uses. What about when you need to go through six hours of video to find the thief? This can make it easier to find scenes in the video where something happened. Or it could be used to delete video scenes where there was no motion. This could be part of an archive procedure to save disk space, only storing video with motion. So we can see that with digital video, motion detection saves money and time. The only question is when do you want to do it? Since we assume all video is recorded, the motion detection can be done during the capture or later during the day or night when there are more computing resources available.
Object tracking, do I need it? Video object tracking, at its basic, is taking a video into a digital format (or converting analog video to digital), and having a computer process the image. This is similar to video motion detection, but more complex as you will see. The computer can look at objects, events, colors, light levels, anything in the video image that could be classified as an “object”. Object tracking is the ability of computers to track that object in the video. Think of a man walking in from of a camera, then walks out of the picture to the left. From watching the video, you know what happened, but with digital object tracking, a computer tracks the objects and can make certain assumptions without you looking at the video. Let’s talk about an application for video object tracking.
In this scenario, we see how object tracking software follows a person as they move into a certain area, which trigger certain events depending on the person’s movements. First we have a guest entrance for “ToonLand”, with a “firearms” sensor that can sense when people walk through a checkpoint with a gun. (these sensors do exist and uses millimeter wave radar to detect large metal objects carried on the person). Here is the sequence of events:
1. The guest entrance senses a firearm. This activated the video object tracking software. The software now tracks the person exiting the gate and notes the time and exit speed and direction of the guest. This information is stored in software. The software triggers another high resolution camera on the moving “object” and puts it on the server for the security force. The picture, showing the upper body of the pistol packing teen, is then sent to all the security guards mobile phones via another layer of software.
2. The next video camera is aiming past the entrance and into a large plaza. This is where we need to “track” the person until he moves into another part of the park. Toonlands operating procedures say we don’t want to make a scene in the plaza. But we do want to “grab” the pistol packing guest. This video software detects that the object moved east into the “Oceanic” area.
3. The east guard is alerted that Mr. pistol packer is heading his way. The waiting guard, hidden from view, sees the man that was shown on the video. The video software already had set an alarm for him, and now he moves and takes the suspect, as the guard in the Oceanic area is the only one able to get him without serious disruption of the pleasantries at ToonLand. The security member asks the guest to come with him, allowing for discussion of the firearm or arrest.
This is just one example of the many uses of object tracking. How else is it used? Object tracking software is used to count the number of cars going on a freeway, following a bad egg in a factory, setting off alarms and tracking targets for the military forces. It is also used for facial recognition, where the software identifies the face in the video and is able to use that for access control or intruder detection.
In this issue we covered object tracking and motion detection. Both are key concepts you need to understand in dealing with digital video and surveillance.
Next month we will look at two different systems for storing digital video, two systems for doing object tracking, and then we will interview two different companies that tailor make video systems using custom programming and off the shelf hardware. We will also take a look at the City of Irvine, a growing city in California. The city has 235 cameras and we will look their roadmap for using digital video. |
|
|
|
|
About the Author |
| Rod is a communication technology expert currently working at a fortune 500 computer company. |
|
|
|
|
|
|