Simple and reliable control of the number of persons within a building with MVTec HALCON

Security & Surveillance | Blob Analysis | Morphology

Strelen Control Systems GmbH is a Certified Integration Partner of MVTec with seat in Büttelborn, close to Darmstadt, Germany. The solutions developed by Strelen are suitable for the inspection and analysis as well as for the automation and regulation of production processes. Their solutions include the development of IIOT (Industrial Internet of Things, aka Industry 4.0) applications and the integration of artificial intelligence in the evaluation of optically obtained data.

The problem

Controlling the number of people in a building can be necessary for many reasons, as the COVID-19 pandemic has shown. The aim of the application is to be able to precisely indicate the amount of people present in a building. Based on this, entry restrictions can be put in place once the maximum permitted quantity is reached. In order to do this, an exact count of each person entering and leaving the building and a calculation of the number of people in the building in real time is required.

The challenges

Safe-Ident People system for controlling the number of people in a building
Strelen's Safe-Ident People system controls the number of people in a building

To ensure reliable and safe counting, even difficult cases must be detected, e.g.:

  • Persons pushing a shopping cart or suitcase
  • Several persons entering the room together
  • Multiple persons moving in different directions at the same time
  • Persons that are running
  • Dogs
  • People in wheelchairs
  • People with headgear
  • People accompanied by children

How it has been solved so far

Before the use of MVTec HALCON different solutions have been utilized: One possibility was to use additional personnel. This method, however, proved to be expensive on a long-term basis and “miscounting” errors occurred. In order to minimize this risk, each customer was obligated to use one of the counted shopping carts. However, additional staff was still necessary to ensure that each customer did use a shopping cart.

Another possibility was the use of widespread marketing analysis systems. However, since these systems do not measure reliably and fast enough, they are not suitable for controlling the number of people in a building.

The solution with MVTec HALCON

Object and position recognition with blob analysis and morphology for outline extraction
Frontend of Strelen Safe-Ident People including HALCON graphic window
Object and position recognition with HALCON

HALCON is used to detect and isolate persons in a depth image. For this purpose, real-time depth images are first acquired with a stereo camera. After some preprocessing, interconnected height profiles are identified and analyzed. Using the mathematical morphological algorithms integrated in HALCON, pixel clusters are segmented, and individual persons are recognized based on their characteristic appearance on the depth image

Using a tracking method, the found persons are assigned waypoints spanning several images to create a path that represents the movement of the identified persons. Finally, the paths found are used to increase or decrease the number of persons.

This is what Strelen says about HALCON

"In the current crisis, the development time for a project is extremely critical, which is why we decided to go with HALCON. The well documented and reliably tested algorithms allow us to concentrate on the important challenges during development".
Jan Strelen, development engineer

Text kindly provided by Strelen Control Systems GmbH.

Find out more about HALCON