Since its inception in the 1980s, machine vision is constantly advancing in optical technology, processors, and software algorithms and making it easier for end-user to use it. so we talk about upcoming trends in machine vision.
Just like in other industries which are benefiting from rapid advancements in technology like big data, cloud, artificial intelligence (AI). Likewise these technologies will also hugely beneficial to manufacturing industries, the three key advances in machine vision for automation will be:
Day by day demand for bulk inspection is increasing which require larger area to be seen by the camera for inspection purpose. 1, 2 and 5 Megapixel cameras continue to lead market but we are also seeing considerable demand of higher resolution sensors such as 12-20 megapixels owing to the requirement of larger inspection area. Earlier multiple cameras were used to cater this requirement but using multiple camera increases a lot of error prone post processing on images which consume huge amount of processing power.
Measuring depth of on object or distance to the object has also turned out to be crucial which was earlier done by using multiple cameras placed at different viewing angles. Now the technology such as stereo vision, time of flight (tof) and leaser triangulation is supported by required advance hardware to meet the requirements.
Traditional methods are still widely used in most food industries due to the unavailability of smart alternatives. Traditional destructive methods are not suitable in today's hypercompetitive marketplace. Consequently, a cost-effective, efficient, rapid, and reliable method is required. In particular, there is a great interest in developing non-destructive optical technologies that have the capability of monitoring in a real-time assessment. Among them, hyper spectral imaging techniques have received ample attention. Hyper spectral imaging systems provide spatial and spectral details; therefore, these systems introduce new sensing facilities that enable improved inspection.
“fabulous machine that have all our senses, all our reasons and think just like we do – you have seen these machine in movies as a friend or foe, now fiction becomes reality”
AI is allowing enterprises to automate inspections that were previously only able to do manually. It efficiently solve complex inspection challenges that are cumbersome or time-consuming to do with traditional rule-based machine vision algorithms.
With deep learning, engineers can train a machine vision system to learn what is an acceptable or unacceptable defect from a data set of reference pictures rather than program the vision system to account for the thousands of defect possibilities.
For example, a pharmaceutical company’s systems
“Human vision has no meaning without a beautiful brain”
“Images from camera has no value without a powerful software”
The Irony is we still don’t know exactly how human vision works, but here is a example illustrating how we humans can do most complicated task with utmost ease.
So we can conclude that brain do well what computers do poorly, and vice versa. Human brain is the best learning machine and can also produce new thoughts / ideas whereas Computers are meant to perform repeated task.
Correlation between human eye and camera can be seen below:
Our understandings of visual information are highly developed. But for computer to understand the images captured by the camera are very difficult. Because we humans visualize an image as a whole while computer software works on pixels (i.e. the smallest unit of image). So below video clip will help us to visualize how difficult it is for humans to work in similar way as a computer dose.
Video 1: https://www.youtube.com/watch?v=NGGLFUuhXDQ
Video 2: https://www.youtube.com/watch?v=NkQ58I53mj
Video 3: Courtesy from Basler
Video 4: Courtesy of BBC