Xovis PC Series, Руководство пользователя

Страница: 89 / 99

89 / 99 www.xovis.com
4 Troubleshooting
False behavior
Under some special conditions, the sensor can show an unwanted behavior. Examples for
such behavior are wrongly detected/generated persons or bubbles of a tracked person which
jumps to an object, e.g. a pillar.
Such issues occur rarely and are caused by a scene situation which disturbs the sensors
algorithm and leads to a false detection. Even if it is unlikely that such false behavior would
influence the counting accuracy of a sensor, it can be desirable to reduce such issues.
Masks
Depending on the actual disturbance, two types of masks can be used: Exclusion and taboo
masks.
As the sensor is robust against disturbances in most situations, masks will only be
necessary in few situations. Proactively drawing masks is therefore not required.
4.2.1 Exclusion masks
Exclusion masks define a zone as invalid for tracking. Same as with the floor mask, the
exclusion mask is oriented on the floor. As soon a person enters an exclusion mask, it will be
deleted immediately. A new person which is located in an exclusion mask won’t be generated
until it leaves the mask. The location thereby is always referenced to a persons’ foot position.
The exclusion mask can help to hinder false detections or bubbles sticking on objects like
pillars, walls or lamps. To identify situations where this mask could be used, it is
recommended to switch the sensor to foot coordinates. That way bubbles are projected to
the floor and potential problematic objects/regions can be identified easily. Refer to chapters
3.2.7.9.1 and 3.2.10.4 to learn how to change the coordinate mode.
A mirroring wall (metal or glass wall) is a common example where exclusion masks can be
used. When a sensor shows problems with bubbles occurring or jumping to the region on the
mirroring wall, the whole wall can simply be covered by an exclusion mask.
Another example for using an exclusion mask could be an information column in the
peoples’ pass way. Especially when located close to the border, the bubbles of persons
leaving the scene could jump to the column and stick there for a short period. To hinder this,
a simple exclusion mask covering the footprint of this column can be drawn.
Figure 53 shows an example for both of these situations. The left image (a) shows a scene
with a big mirroring wall on the left and an information column on the right bottom. The right
image (b) shows the same scene with exclusion masks applied.