Machine vision is a broad term with many applications and definitions, but at its most basic definition it is using an imager or camera along with software to inspect and analyze a real world object. In addition to the camera and software, machine vision systems usually include lighting, optics, and pass/fail mechanisms.
In a strong machine vision application the camera takes images of the object under investigation. Proper lighting and optics create a clear image or images that can be analyzed by the software specific to that application, which can then provide a pass/fail response.
Cameras used in machine vision applications come in a myriad of varieties. Sensor size, resolution, and sensitivity to light at different wavelengths all play a part in how well the camera can inspect the material. The camera selected must meet the specific needs of the application where it is to be used.
Lighting can be used to increase the camera's ability to see the product. The light source's power in lumens, color, wavelength, and positioning all play a role in enhancing the image that the camera receives. The wavelength of the light can provide additional information on the material. Different wavelengths have different transmissivity and can be used to enhance the texture of the material in the image. The wavelength of the light controls color which also plays a role in enhancing the quality of the image. High contrast light in reference to the material creates further clarity for edge detection and imaging.
Optics are used to bring the part into focus and to make the image available at the full resolution of the camera. The correct lens with the appropriate characteristics, such as magnification and focal length, must be implemented to make full use of the camera.
Under these ideal circumstances, the camera passes the information to software specifically designed to fit the application. This software is equipped with a pass/fail mechanism, which is the end result of the application. It qualifies the material under investigation, and in some cases can trigger a response. In the case of a bottling plant, it could trigger a reaction that rejects and removes the non-qualifying material from the production line. In any scenario, machine vision applications are designed to improve quality and save time and money. They can perform tasks faster and with more accuracy than what can be accomplished with human interaction.
Machine vision is typically associated with production and manufacturing industries, but its reach extends so much farther: from medicine to agriculture, traffic monitoring, security, and even laser measurement. Just as machine vision can be used to find imperfections and to qualify manufactured parts in production, it can and is used to find imperfections and qualify lasers used in a variety of fields.
For many years now machine vision has been used in laser measurement applications, even if not recognized as such. High-resolution cameras and powerful software are being used to image and analyze laser beams, replacing dangerous and inaccurate methods that used burning. Cameras of varying resolution are selected based on their ability to see into different fields of the light spectrum. The camera's ability to see a laser beam is directly affected by its sensitivity in certain regions of the spectrum. Beam images are taken and analyzed to give clearer information of laser beam behavior. These camera-based systems have created a new standard for the ways that beams are measured and qualified.
Before the use of cameras, high power lasers were measured and tested using burning methods. A beam would be directed to burn paper or plastic. Such methods can be dangerous and only provide information about the width and shape of the beam. Through machine vision methods, measuring lasers has become much safer, faster, and more informative.
Traditionally, laser properties have been measured through direct contact with the sensor of a high-resolution camera. This means that the laser beam is placed directly on the camera's delicate sensor. While this can provide much more detail, it is not without its limits. High power laser beams pose a particular challenge where such beams have a tendency to damage any sensor put in their path. Advances in the laser measurement industry have been implemented to address this issue. The most basic of these advances are beam attenuators and splitters. The issue remains that not all beams can be analyzed through direct contact. For the beams that are too large or so powerful that they would damage the sensor and attenuators, different methods must be used.
One method to address this issue is through indirect imaging of the beam. There are three variations of this method. The first method, shown in Figure 1, has the laser being directed onto a surface that is spatially consistent and can take the power of the beam without being burned or damaged by the laser beam. This surface is referenced here as the projection surface. The imaging camera is then placed at an angle to the projection surface.
The second variation is the opposite. The camera is placed looking directly at the projection surface and the laser is placed off axis. In both of these methods, the image is sent to computer software where it is displayed and analyzed.
The third indirect method is referred to as transmissive imaging. It is similar to the two methods already mentioned, except the beam is cast onto a transmissive optic. The camera is then placed on the opposite side of the transmissive optic at a slight angle so that the beam does not contact the camera's sensor directly. The camera then sends the information to the software on the computer where it is displayed and analyzed.
But not all lasers can be measured with these methods. The high-power lasers found in industrial and additive manufacturing are too powerful for even the indirect methods discussed above. One technique to overcome this difficulty is to place the sensor in the beam for only a short period of time, then cover the sensor, typically with a physical shuttering system, before the sensor becomes damaged. This works well as long as the laser is not powerful enough to damage the shutter and the sensor is protected long enough to recover before being exposed again. A newer method to measure these beams is with a completely non-contact system. Systems like the Ophir-Spiricon® BeamWatch®and BeamWatch® AM are designed to address this need.
Such non-contact systems are a good example of using all the features of a true machine vision system. Industrial lasers often operate at an NIR wavelength (around 1060nm-1080nm), which is invisible to the naked eye. Therefore, camera sensors are required that are sensitive in those wavelengths. In non-contact measurement, the Rayleigh scattering is imaged and the beam caustic is measured by the software installed on the computer. What makes this a difficult task is that Rayleigh scatter is inversely proportional to the wavelength to the fourth power. This means that not only are typical cameras less sensitive in NIR wavelengths but the scatter is significantly reduced, which further decreases the camera's ability to see the beam. A dark background enhances the image of the beam, which is housed in a dark container with coatings that prevents the beam from further reflecting and distorting the image. Purge gas is also used to create a clean, pristine environment in which to see the beam. Beams can be seen as low as 1.5 MW/cm2 up to intensities we're yet to see in the lasing world.
Once images are taken and enhanced to provide full detail, they are processed by software that calculates parameters such as focus spot size, focus shift, divergence, and other qualifying parameters. An advantage of using the Rayleigh scatter technique is the ability for users to see beam focal shift, a characteristic that was previously difficult to measure using any method. As the optics inside a laser delivery head heat up, the focus position changes. This becomes even more of an issue as the optics get dirty or damaged. Users are now able to, in real time, make a determination on the quality of the beam and decide if service or adjustment is required.
These advances in machine vision techniques used in laser measurement systems have enhanced the way that laser beams are measured and qualified in numerous laser industries. Thanks to the use of cameras and sophisticated software, lasers can be imaged and measured safer and more accurately than ever before. The future of machine vision in laser measurement systems certainly looks bright.
This article was written by Chris Jones, Technical Writer, Ophir (North Andover, MA). For more information, visit here.