NI Machine Vision Provides a Cost-Effective Plastic Sorting System


"To perform plastic sorting at lower cost, we selected NI hardware and software tools to ensure quick testing, efficiency, and robustness of the solution. "

- Sebastien Parent, Averna Technologies, Inc.

The Challenge:
Developing an automated sorting system for clear transparent PETE plastic with machine vision technology – instead of the standard infrared spectrometry or x-ray equipment –performed at a fast rate of 40 parts per second.

The Solution:
Combining a special lighting technique, a specific camera acquisition mode, the processing of image in color space, and a well synchronized rejection device.

Sebastien Parent - Averna Technologies, Inc.
Francois Vachon - Averna Technologies, Inc.

The plastic recycling industry needs to precisely separate the plastic by category. The purity of each plastic group (PETE, PVC, PP, etc) has a direct correlation to the refund money. Because the product is not consistent throughout the manufacturing industry, manual sorting tasks are performed, but there is a real automation need.

The existing techniques for plastic sorting include using x-rays and infrared spectrometers. The advantages of these techniques are to determine the nature of the plastic by the signature in the various wavelengths at a molecular level. But the technology is complex and expensive. The advantage of using machine vision in the visible light domain is a low cost with the capability of using standard machine vision algorithms.

To perform plastic sorting at lower cost, we selected NI hardware and software tools to ensure quick testing, efficiency, and robustness of the solution. The plastic parts to sort come from an in-feed conveyor that drop in front of the camera looking at a specially designed backlight. As long as no plastic, labels, or opaque plastic block the lighting, the camera sees no information. But as soon as a translucent plastic is between the camera and the lighting, it is detected by the inspection system and processed for sorting. When the system detects PETE colorless plastic, air nozzles fire to change the path of the plastic in a recycling bin. For all the other material type (colored PETE, opaque plastic, paper label, plastic labels, plastic other than PETE), the plastic parts fall on an output conveyor without any disturbance of the drop path.

In this application, the major challenge is to perform all required tasks in the amount of time available to ensure 100 percent plastic inspection. The conveyor speed is 4.5 feet per second, but the speed of the plastic part is around 9 feet per second, considering the acceleration caused by the gravity. A field of view of 2.8 inches in height per 28 inches long with a resolution of 0.044 inch per pixel, 40 images of 64 x 640 pixels has to be processed. At the end of each process, the system applies rules to determine which air nozzle to open.

To reach the image acquisition rate with the selected field of view, we selected a CMOS color camera from PixeLINK. The camera is used in a mode where the frame rate is increased by windowing of the sensor.

The system is designed to keep a fail-safe logic. In our specific application, the goal was to have the PETE non-color product 100 percent pure in the recycling bin. More tolerance is given on the color and non PETE product because it is sorted manually on the following conveyor.

The strategy to perform a quick and efficient development was to first use Vision Builder as a prototyping and functionality testing platform for processing time evaluation and selection of algorithms. We also used Vision Assistant software for filtering tests. As soon as the general strategy was determined, we translated the Vision Builder sequence in NI LabVIEW VI. This was the starting point of the system solution. We then proceeded with optimization and system completion using LabVIEW.

We were challenged with having the maximum of image processing tools in the 25 ms available time for each image. With Vision Builder, we determine the analysis strategy in four major steps:

1)      Image acquisition,

2)      Thresholding,

3)      Pixel count, and

4)      Logic for rejection.

1. Image acquisition

Using the windowing functionality of the camera and an exposure time of 1 ms, the acquisition time is 2.3 ms per image. A typical image is shown in Figure 2.

2. Thresholding

The second operation is to perform two thresholds on the acquired color image. A first threshold is to keep only the white information. Another threshold is to extract all color information. To perform these thresholds, we work in the HIS color space, which gives a more realistic view of the product variation. The white information could be darker or clearer and still be considered PETE non-color plastic. The processing time of each of these two thresholds is 7.1 ms.

3. Pixel count

Once the white and color information is extracted from the initial image, the system performs a compilation by rejection zone. The two thresholded images are then separated in 20 equal areas of 64 pixels height by 32 pixels width (2.8 inches by 1.4 inches areas). The operation is to perform a pixel count for each of these nozzle areas. To save processing time, the system can open all image buffers at the initialization stage to save the time allocated for this task. To implement this feature, we made a counting pixel algorithm with LabVIEW instead of using the Count Pixel VI from Vision Builder. We then gain a factor two in processing speed (0.148 ms instead of 0.270 ms).

4. Logic for rejection

Using the result of pixel summation for the two images, a comparison-logic software determines the air valves to activate. This code keeps a fail-safe strategy for the plastic sorting. The air nozzles are kept closed around color plastic even if PETE non-color plastic is present.

The response of the air nozzle is faster and the rejection equipment to the visual inspection area is closer. We then minimize the risk of plastic position variation during falling path, which can cause product mixing. Even if the I/Os activation is performed in parallel with the program, to ensure an optimum activation speed, we perform a port output access instead of a line output access.

The solution uses LabVIEW image acquisition algorithms to optimize the performance in the available processing time limitation. The advantages were the fast development procedure with Vision Builder and the flexibility in LabVIEW for optimizing the various functions of the system and to complete the entire system with all needed inputs and outputs. The final result is at half the price of any standard solutions.

Author Information:
Sebastien Parent
Averna Technologies, Inc.
700 Wellington, suite 1400
Tel: 514-843-7577

Bookmark and Share

Explore the NI Developer Community

Discover and collaborate on the latest example code and tutorials with a worldwide community of engineers and scientists.

‌Check‌ out‌ the‌ NI‌ Community

Who is National Instruments?

National Instruments provides a graphical system design platform for test, control, and embedded design applications that is transforming the way engineers and scientists design, prototype, and deploy systems.

‌Learn‌ more‌ about‌ NI