+7 8202 20-16-35
Mon-Fri 9:00AM – 6:00PM (MSK)
Array
(
    [0] => Array
        (
            [value] => 650
        )

    [1] => Array
        (
            [value] => 1187
        )

    [2] => Array
        (
            [value] => 721
        )

    [3] => Array
        (
            [value] => 653
        )

    [4] => Array
        (
            [value] => 681
        )

    [5] => Array
        (
            [value] => 680
        )

)

Product control in production: machine vision cameras or IP cameras?

IP cameras have proven themselves in security applications, but can they be just as successful for quality control and product tracking in industrial environments? They are easy to use and much cheaper than specialized machine vision cameras, giving the mistaken belief that they are competitive. But is this really the case? What is the difference between machine vision cameras and IP cameras and will the savings on equipment be justified in the long term?

Product control at the production facilities that value their name and reputation is carried out with the help of computer vision, which allows to increase the accuracy, speed and efficiency of information processing, eliminate the human factor and minimize the probability of error. Control automation is particularly effective when the volume, speed, or complexity of the information to be analyzed significantly exceeds the ability of the operator.

The advantages of using computer vision are obvious, the only question is the choice of its components to solve certain tasks. First of all, it concerns video cameras.

Today, there are many offers on the market in different price categories for different purposes. At the same time, it is widely believed that IP cameras can cope with the tasks set by industrial production — such as defect detection, surface inspection, identification of codes and markings, assembly inspection, product identification and classification, and many others. But can they? Let's get to the bottom of it.

The conditions and tasks of companies in various industries, from pharmaceuticals to electronics and metallurgy, place the following requirements on video cameras:

  • High sensitivity of the sensor and high quality of the obtained images with a resolution of at least 2448×2048 pixels.
  • Supports standardized GigE/10GigE/USB 3.0/Camera Link/CoaXPress software interfaces.
  • Possibility of generating frames based on an external signal, which is important for analyzing individual products in production.
  • Ability to operate in real time. Requirements for the maximum acceptable response time between trigger signal and image acquisition can range from microseconds to seconds.
  • Possibility to receive synchronized images of the monitored item from several video cameras.
  • Water and dust protection, wide operating temperature range and vibration resistance.
  • High quality of assembly and reliability.
  • Availability of digital lines for connecting external devices.
  • SDK availability — development tools for various programming languages.

Within these requirements, the advantages of machine vision cameras over IP cameras are obvious:

  • Machine vision surveillance cameras are designed and constructed to solve optical inspection tasks using software and allow for fine-tuning adjustments to the resulting image, such as shutter speed, saturation, etc., depending on the task at hand. IP cameras are designed mainly for video surveillance, which does not require high definition of the picture, which is essential for industrial monitoring. The processing of images from IP cameras in most cases occurs during visual control by a human, and to integrate IP cameras into specialized software will require additional costs.
  • Due to the lack of data compression, machine vision cameras provide raw images of higher quality suitable for correct operation of computer vision algorithms. This approach allows preserving the smallest details of the original image, which is indispensable, for example, in surface inspection or dimensional measurement systems. Frames from IP cameras are processed by a certain codec, i.e. compressed, which leads to loss of information. If the flaw is located in the part of the image for which the data is distorted by compression, the quality control system will erroneously pass the flawed component as compliant with the quality standard, i.e. will fail in its task.

The figure shows images of the rock mass on the conveyor belt taken from an IP camera (left) and a machine vision camera (right) in motion (top row) and static (bottom row):

  • Machine vision cameras can have a resolution of up to 86 MP for area scanning or 4K for line scanning — and still capture up to 200 frames per second. The frame rate at the maximum resolution of IP cameras is limited to 20 frames per second on average, and the delay due to video stream compression by codecs an be up to 1-3 seconds. Such delays are unacceptable when analyzing fast processes, for example, when the inspected products are moving along the conveyor belt at high speeds.
  • Machine vision cameras, unlike IP cameras, allow you to set up precise frame synchronization. Such functionality is necessary, for example, when reconstructing 3D diamond shapes in free fall and other similar tasks.
  • Another feature of machine vision cameras is the ability to select the optimal components, such as lens or lighting, which allows for flexibility in selecting the equipment for a specific task and getting the most effective result. Typically, IP cameras are the out-of-the-box product, which does not imply any modifications.

Thus, in conditions of insufficient accuracy and resolution of images received from IP cameras, the probability of error increases, which is unacceptable when solving the tasks of quality control and production process management by analyzing visual information. In the long term, such errors lead to multi-million dollar losses due to spoilage, product identification failures, and insufficient measurement accuracy.

IP cameras can be successfully used for monitoring of large-sized objects, but as soon as a detailed inspection is required, machine vision cameras are the only right choice.

Machine vision cameras require a little more investment during the installation phase of the system, but they quickly pay for themselves by providing a reliable solution for production tasks. For the company, this leads to benefits such as improved product quality, increased labor productivity, reduced volume of defected products and related costs, reliable product monitoring, and a positive reputation among customers.

As a developer of industrial control solutions for a variety of industries, we use specialized machine vision equipment from the world's leading manufacturers as part of the systems we supply. The high- quality video stream, multiple functions for a wide range of applications, industrial design and a wide range of lenses and accessories provided by these cameras give us confidence that our customers will receive a reliable solution, meeting their requirements to the maximum extent possible.

Obviously, the selection of machine vision equipment is not a simple task, which requires high competence and experience of implementation in industrial production, for that reason it should be entrusted to professionals. In this case, the implementation of machine vision system at the production site will result in the positive user experience and achievement of the set goals.

Written by Nina Aksenovskaya
Published in November 2024 issue of  «Control Engineering Russia» journal

20.11.2024