phil-vision: Why Phone Cameras are Insufficient for Machine Vision
From quality control in the industry to autonomous navigation of self-driving cars, machine vision is used in a wide variety of applications to automatically analyse, inspect and measure objects. At the same time, as cameras in mobile phones are getting better and better, we are often confronted with the idea of using mobile phones for inspection tasks as well.
However, when it comes to machine vision, there are several reasons why a mobile phone is not sufficient for industrial applications. Sure, developments in mobile phones are rapid and there are new phones with even higher resolution sensors every year, but it often takes years before these are available in the much smaller machine vision market. Also, the development of a machine often takes several years and by the time the system is ready for the market, the mobile phone used is obsolete. But availability is not the only problem; a number of other factors speak against the use of mobile phones in machine vision.
The right illumination is crucial
Proper illumination is essential for machine vision applications because the amount and quality of light has a massive impact on the accuracy of the results obtained by machine vision systems. For example, if an object appears too dark or too bright due to inadequate lighting, it is usually not possible to accurately recognise the object’s features or accurately measure its size or position.
Using a mobile phone with built-in illumination is therefore extremely challenging. The mobile phone’s lighting can only emit a certain amount of light, so it is often not possible to achieve the desired illuminance. It is also difficult to ensure that the light intensity does not vary across the entire field of view. The use of an external illumination with a mobile phone is also difficult because the camera of the mobile phone cannot be synchronised with external lighting and the use of continuous lighting is only possible to a limited extent.
Importance of a repeatable and robust set-up
A reproducible, robust setup is essential for vision applications. The environment in which the vision system operates must be consistent and reliable in order to produce meaningful results.
Using a mobile phone as part of a vision system is therefore extremely challenging, especially when it comes to achieving repeatable and resilient results. Since mobile phones are sensitive to their environment and can be easily affected by changes in temperature, humidity and other external factors, a robust test system setup is impossible. In addition, mobile phones can be easily moved or shifted due to their size and weight, further complicating the setup of a repeatable and robust testing scenario.
The fact that a mobile phone is guided by “hand” directly to the defect and held in different directions until a good contrast of the defect is obtained before a high-resolution image is taken where the defect is almost image-filling, produces a great snapshot, but makes a reproducible set-up impossible and is not applicable even for slower production processes.
Inspecting moving objects
Mobile phone cameras are exclusively equipped with rolling shutter sensors. However, unless these sensors are specifically flashed, they are generally not suitable for measuring tasks when objects move quickly, because the recording times between the first and last line are different. In machine vision, global shutter sensors are used for this purpose, which is not available in mobile phones.
Different lenses / focal lengths according to object size and field of view
Depending on the size of an object or the field of view, different focal lengths and lenses are needed in vision applications. For example, if you want to inspect a large object, you need a longer focal length to capture it in its entirety. On the other hand, if you are working with a small object, you need a shorter focal length and a macro lens to see small details on the parts you are inspecting.
Since most mobile phones only have a fixed focal length, this limits the applications considerably. And even if you manage to attach different lenses to your mobile phone, there is no guarantee that they will work properly as they may not be compatible with the phone’s camera sensor.
Industrial machine vision solutions must be stable, and the components used should be easily replaceable even years later. This is impossible with mobile phone-based solutions because a mobile phone bought today is mostly already obsolete soon.
Probably the most important distinguishing criterion is the fact that industrial cameras and consumer cameras are designed for completely different purposes and should be used as such. With mobile phone cameras, it is important that the images look perfect to the human eye and a lot of effort is put into post-processing so that edges, contrasts and colours look as realistic as possible to the viewer. With industrial cameras, on the other hand, the image often appears “dull” to the viewer but is ideal for measurements.
However, customers often expect an image quality similar to that known from mobile phone cameras. However, achieving this image quality is only possible at considerably higher costs because the optimisation and pre-processing functions that are necessary to enhance, de-noise, de-shake or sharpen images or to correct colours often have to be redeveloped individually and at a high cost.
Therefore, machine vision often has to work with lower-resolution images, but these disadvantages can be overcome, for example, by using multiple cameras or dedicated illumination. The robust design and long-term availability also make the use of industrial cameras for 24//7/365 machine vision applications in industrial environments without alternatives.
Find out more about phil-vision.
Also, stay up to date with the most recent machine vision and image processing news right here on MVPro Media.
MVPro Media’s Matt Williams spoke to Georgy Das from MidOpt (Midwest Optical Systems). Das explained...
Zebra will present warehouse solutions built on Industry 5.0 human-centred automation at IntraLogisteX. Warehouse operators...
To support its customers in the critical task of selecting the optimal line scan camera...
Additive manufacturing, or 3D printing, allows Seco Tools to create products that would otherwise be...
In many plants you will find cameras, lighting and other hardware that was once part...
In the food industry, it’s essential to carefully control the temperature of perishable goods throughout...
The new Mech-Eye DEEP 3D camera is the perfect solution for logistics applications. Combining fast...