Setting the Size of a Person
There are three settings governing the size of a person in the picture: the minimum width, the typical width and the typical area.
A person will not be counted unless they exceed the minimum width.
The typical width is used to decide if an object is one or two people.
The area of objects crossing the counting zone is continuously variable. The pixel "blob" may be one person, two people side-by-side, one person following another, a person pushing a pram, etc. Two people side-by-side would be a wide, thin are. Two people following one another would be a long narrow area. The typical area is used with the typical width to decide how many people the object represents.
- Select the VT Module
- In the Settings list, alter the MinDetect, Typical Width and Typical Area. You might like to start with the default settings and only make changes when you are testing the system
Minimum Person Size (MinDetect)- Set the minimum person size to match the minimum size of an adult or child to be counted.
- If the minimum size is too small then small objects and animals may cause counts.
- If the minimum size is too big then children, small adults or adults wearing clothes that closely match the background colour may not be counted.
Typical Person Width- This should be bigger than the minimum person size (MinDetect).
Typical Area- Where the camera gives a view of a good-sized area, people have typical areas of 40-80 pixels.
- If the camera is mounted on a low ceiling, people will look very large in the picture and their typical areas will be bigger.
- Click the Apply to VT button or Send Settings to Module.
- If you wish to set other VT units to this setting go to the Settings tab and click button 5, Copy settings to higher address numbers.
Note: changing the height of the camera above the floor in software will also alter the size of the person detected. Between 2 and 3 metres above the floor, the area of screen taken up by a 1.8 m tall person changes considerably. A person viewed in a camera 2 metres above the floor will appear around 3 times as large as when the camera is 3 metres above the floor.
Manual ref: VT-2.04, June 2007 |