ENJOY THE STYLE AND THE ACCURACY OF COUNTING

How It Works
3D people counter employs a stereovision algorithm that allows counting dense traffic with high accuracy.
The DSP-processor receives the picture from two graphic sensors, calculates the tridimensional image and turns it into a mathematical 3D model, determines where the heads of the customers are and distinguishes them using the embedded algorithm. The processor tracks the customers within the coverage area, discounting false objects, shadows, specks of light, shopping trolleys, bags, children and strollers.

The processor checks each point in the image from the left camera and identifies the matching point on the image from the second camera.

t - perceptible point of the customer’s head
t' - projection of t on the left camera
t'' - projection of t on the right camera

The 3D DSP-processor calculates the relative shift of the t' and t'' points.

Circuit Design
CMOS graphic sensors receive the picture of the coverage area through two lenses. The sensors digitize the primary image and transfer it to the DSP-processor, so they transfer the same image (but shifted along the base line) at the same time.
The DSP-processor receives the images, adjusts the colour, brightness, noise, and buffers it. It compares the two edited images, searches for matching details and based on the result calculates and creates a mathematical 3D model of the location. Using the installed software, the processor determines the positions of customers’ heads and counts how many of these people went through.

Algorithm Visualization
The videos show a 3D stream of people without any shadows. The algorithm calculates the height of each person.

Parameter Name	Parameter Value
Type	people
Technology	camera