In this article, we’ll be taking a look at what 3D machine vision is and how it works, as well as going through a few different applications these machine vision systems have. In order to understand the benefits of using 3D machine vision in the applications listed a little further down, we’ll look at the different ways in which 3D machine vision is achieved how these varying methods affect the applications the system is best suited for. As we’ll see, many of the applications of machine vision can be vastly different from each other, and many are often not particularly obvious on first consideration. However, we can be fairly certain that, as we develop machine vision and more applications for the technology become apparent, these systems will begin to feature more and more in our everyday lives.
What Is 3D Machine Vision?
So what is 3D machine vision? And how exactly does it work? 3D machine vision can be roughly defined as technologies that allow for the three-dimensional measurement or inspection of 3D objects or surfaces. There are a few different ways in which this is achieved:
Laser Profiling: Laser profiling is one of the most popular 3D imaging techniques. The object being measured moves through a laser beam as a camera positioned at a known angle records the changing profile of the laser as the object moves through it. This configurations is particularly popular on factory production floors or packing lines as it relies on the object moving relative to the laser, meaning it is highly suited for products on conveyor belts.
Stereo Imaging: Another popular 3D imaging technique is stereo imaging, where two cameras are used to record 2D images of an object that can then be triangulated and made into a 3D image. Like laser profiling, this techniques also allows for the movement of objects when measuring and recording. Using a random static illumination pattern can also give arbitrary texture to plain surfaces and object that don’t have natural edges, which many stereo reconstruction algorithms need.
Fringe Projection: In fringe projection, a stripe pattern is projected onto the entire surface area to be measured. The image is then recorded by a camera positioned perpendicular to the object being measured. The point cloud created is able to give height resolution up to two orders of magnitude greater than a laser profiling method is able to provide. Fringe projection is also more scalable with a measuring area that ranges from one millimeter to over one meter.
Time of Flight: The time of flight method measures the time a light pulse takes to reach the object being measured and then return. The time taken to measure each image point will vary depending on the object’s size and depth and thus each point will provide this information as they are measured.
Applications of 3D Machine Vision
Now that we’ve looked into the various ways in which 3D images are calculated, we can start to look into the various applications of 3D machine vision in different roles and industries. As previously mentioned, machine vision is no new concept and has established itself as an essential technology for the modern age. In recent years, however, technological advancement and the cheapening of 3D imaging systems has meant increased adoption and development of 3D machine vision and with that has come the implementation of these systems across many industries in a variety of roles. Let’s now take a look at a few of those applications.
Robotic Guidance and Automation: One of the most obvious applications for 3D machine vision is in robotic guidance and automation. Car manufacturers, for example, use armies of automated robotic arms to build their parts and vehicles. A three-dimensional image allows these robots to perform ever more accurately and handle precise manufacturing calculations and movements when it comes to constructing parts or vehicles. Usually, these robots are sealed away from human workers due to health and safety procedures, however, 3D machine vision systems could eventually allow robots to detect a human presence and calculate the safest movements around them.
Quality Control & Inspection: Quality control (QC) and inspection is another application of 3D machine learning many organizations are taking advantage of. Vision equipped machines are able to identify anomalies or defects in products or objects and flag them as not meeting certain quality assurance specifications. 3D machine vision could also be applied to vehicles and help booster predictive maintenance efforts. A good example of 3D machine vision being used for inspection and QC would be that of a university in Belgium where they developed a 3D machine vision system that was used to detect defects in harvested apples.
Mapping: 3D mapping is used in a variety of different industries. From construction and manufacturing to geology and astronomy, 3D mapping is helping us not only build better homes and products, but also uncover the secrets of our unmapped oceans and the celestial neighbors we both have and have not visited. In the present day, 3D mapping is being used mainly to help automation and machine management systems collect spatial and depth information about an area or object, however, in the future it could be used by autonomous robots to map unexplored terrains or space that could then be converted into a highly detailed 3D map for human explorers.
These are some examples of how 3D machine vision technology is currently being used, as well as a few possible future applications. What is worth noting about the growth of 3D machine vision is that it hasn’t been a quick explosion of innovation that caused these most recent of developments, but rather, the advancement of technology over time providing new opportunities and applications for use as these technologies mature and are refined. With the emergence of the Internet of Things (IoT) and the slow but steady progress with artificial intelligence (AI) soon to make other technological innovations such as driverless cars a reality, we can be fairly certain that 3D machine vision is as yet in its infancy. What factors and environments will shape the future development of 3D machine vision we can only guess, however, it seems likely, given the predicted increase in human-machine interactivity, that we’ll be using these systems well into the future.