Machine Vision Cameras for Waste Sorting & Recycling
Advanced technology, such as machine vision, is playing an increasingly crucial role in improving how we sort and recycle waste. A vision system with machine vision cameras makes recycling faster and more accurate. Most often, these cameras are installed above conveyor belts in recycling facilities. As waste moves along the belt, the cameras take high-quality pictures of the items. Using computer vision software, the machine vision cameras can quickly recognize different materials like plastics, metals, and paper. This automated process helps to sort the waste correctly, reducing mistakes and saving time and money.

Table of contents
One of VA Imaging’s customers reached out for help in selecting a suitable machine
vision camera setup for waste sorting. One of our vision experts advised the recycling company on a complete setup based on the specifications of their application. This article describes in only 5 steps how all vision components for the waste sorting & recycling system were selected. Would you like to receive personal advice for your application? Simply share some specifications of your project below and we will reach out to you.
Camera selection for waste sorting system
Advising the right setup for the waste sorting application starts with selecting a suitable machine vision camera. The vision expert of VA Imaging recommended using our 16MP camera ‘ME2S-1610-24U3C’ with the Sony IMX542 sensor because
of the following requirements requested by the customer.
Firstly, an USB3-interface was recommended as the camera interface. This was based on the fact that the customer could benefit from the four times higher bandwidth compared to a Gigabit Ethernet (GigE) camera. In addition, the USB3 interface is recommended for applications where the distance between the pc and the camera is not more than 4.6 meters because of the limited cable length. In the customer’s application, this length was approximately 4 meters. Furthermore, different colors of garbage had to be detected. That is why a color camera was advised.
Additionally, the choice was made to use a global shutter camera instead of a rolling shutter camera. Global shutter cameras are used in applications where either the camera or the object is moving while the image is being captured. In this case, the conveyor belt is moving and does not stop during image capture. A rolling shutter camera, which works great for stationary applications, is not suitable for this purpose. Using a rolling shutter camera would result in a lot of image distortion due to the movement. All cans, bottles and other garbage on the conveyor belt may appear ‘leaned’ or ‘tilted’. In industrial applications such as waste sorting, this distortion leads to inaccurate detection because some waste may not appear as it actually is. A global shutter camera prevents these errors by capturing realistic, undistorted images. More info about Global shutter vs. Rolling shutter can be found in this article in our Knowledge Center.

The customer did not know what resolution was required to recognize the different types of garbage. That is why our machine vision expert offered to help calculate this based on two specifications: the required field of view (FOV) in mm and system resolution (mm/pixel). The conveyor belt on which the waste moves has a width of 425mm that should be shown. The height was not important. A system resolution of at least 0,1mm/pixel was required.
Based on this information, a machine vision camera with at least 4250 pixels wide had to be selected. Our 16MP camera turned out to be a great solution. This camera has 5320x3032 pixels, resulting in a system resolution of even 0,08mm/pixel. The calculation is shown in Figure 1, the resolution calculation created in Microsoft Excel.
In conclusion, this 16MP color camera with the IMX542 sensor offers an affordable machine vision camera solution for waste sorting and recycling while benefiting from the high resolution and light sensitivity.

Selection of lens for IMX542 camera
For the 16MP vision camera to work, VA Imaging’s vision expert selected an appropriate C-mount lensfor the camera. The right focal length lens was calculated by the vision expert based on again two specifications: the field of view (FOV) in mm and working distance in mm. The working distance is the distance from the object to the camera. In addition, it is important to select a lens that is suitable for the sensor size of the 16MP camera. The 16MP sensor. camera has an image sensor size of 1.1”, the Sony IMX542
To calculate the most suitable C-mount lens for this machine vision camera, VA Imaging provides its own online Lens calculator tool. Based on the camera resolution, required field of view and working distance, the focal length could be calculated. The customer indicated that he wanted to see at least 425mm width at a working distance between 950 and 1350mm.

Our 35MM C-mount lens ‘VA-LCM-25MP-35MM-F2.4-120’ turned out to be a great lens for the 16MP camera. Using our 35MM lens on a working distance of approximately 1056mm results in a field of view of 425x242mm. This is shown in Figure 2, the lens focal length calculation.

Accessories for waste sorting system
Machine vision cameras, such as the 16MP camera, should be connected to a pc. To ensure a stable connection between this pc and the machine vision camera, the machine vision expert advised using one of VA Imaging’s USB3 cables. For this waste sorting vision system, our 4.6meter USB3 cable was recommended based on the fact that the length between the pc of the customer and camera is 4meters.
In addition, our tripod mounting plate for this IMX542 sensor camera was selected by the vison expert for easy integration of the waste recognition system.
Machine vision lighting for waste recognition
Adding machine vision lighting to a vision system like this wastesorting application is highly recommended. The customer told VA Imaging’s machine vision expert that the factory environment where the setup will be placed was not well illuminated. That is why they decided to put a closed-off box around the conveyor belt where the waste will be detected on.
The products on the conveyor belt are all very different. From cans to bottles, in all kinds of materials and colors, they should be detected for correct waste sorting and recycling afterwards. To illuminate all garbage as evenly as possible and with minimal reflection, it is recommended to use 2 of our LED spotlights: Industrial LED Spot light VA SL-90x80-W. One of the biggest advantages for this waste sorting system is that the LED spot lights come with three optical lenses. By choosing the right optical lens (30,60 or 90 degrees), they can be placed in the optical angle that gives the most ideal light for the application. It offers a lot of flexibility for the customer.
To exclude all ambient light in the factory, our machine vision expert also recommended building a box around the conveyor belt. The advice is to make the box, for example by painting, completely white colored from the inside. Then place the LED spots on two inner sides of the box and position them upwards. This way, the light 'bounces' to the ceiling of the box and down onto the waste that is on the conveyor belt. The light will be evenly distributed.
To power the two LED lights, our 24V 150W power supply was recommended. This power supply works as a light controller, allowing the LED spots to be triggered and dimmed. In addition, extension cables can be selected on our website if desired.

Image processing software for waste recognition
This specific customer had no experience yet in selecting hardware for a vision waste sorting system. However, their company was very experienced in writing and implementing software. For this specific machine vision application of waste sorting, the customer preferred writing their own software in Python. They decided to consult our free SDK for example programs, including the Python sample to acquire images. One of the articles in our Knowledge Center provides more information about using a Python sample using PyCharm.
For the initial programming, our free Software Development Kit (SDK) was used to capture and save images and set the camera parameters for the waste sorting. The SDK supports operating systems including Windows, Linux and Android and is compatible with regular and industrial PCs and ARM platforms. The SDK supports programming languages like C++, C#/.NET, and for this vision system: Python.
Support for machine vision applications
Would you like to have support from one of our machine vision experts for the best machine vision solution for waste sorting? Or another similar recycling or garbage application? Don’t hesitate to reach out by using the form via this link: Support for Vision Application.