Each year, in cooperation with the trade journals Vision Systems Design and Inspect, FRAMOS – the image processing specialist – identifies trends in the industry for technical and economic development based on users’ and manufacturers’ views. For the 9th market survey, 60 manufacturers and users from 19 countries provided answers about the status quo of the image processing market and its further development.
The current focus of image processing is on automation for the smart factory and science. Embedding of systems and intuitive implementation support the unbroken willingness for investment. Stable prices and strong brands as well as technological advances in the area of sensors and interfaces drive the robust Embedded Vision. Miniaturization and intelligent use of data in networked processes support scalability and profitability, even for medium-sized enterprises. Augmented Reality and Deep Learning are seen as added value and strategic areas of growth for imaging technologies within the IoT and the industry digitalization.
The survey, taking an average of 15 minutes, is based on the statements of 52 users and 8 manufacturers who gave comprehensive information about cameras, sensors and applications as well as a forecast. At 55%, European participants are the largest group, America is represented with 23% and Asia/The Middle East with 22%. Relevance ranking was carried out based on the queried purchase or production volumes. Purchase and production can mainly be found in Europe, at 62% and 43%. Manufacturers also produce in Asia (13%) and in America (6%). Behind Europe, Asia and America act as equally strong purchasing markets, with 28% of users each. Compared with 2015, there has been a decline in the American production market only (28%), which can be ascribed to weaker survey participation from North and South America.
Industry 4.0 and science are currently the focus of image processing
Based on opto-sensory measuring technology and quality assurance, production automation and robotics are the main areas of application for image processing for manufacturers and users. Between 40% and 63% of all participating companies stated the use of visual sensor technology in these areas. Making up about a third, science applications in medical technology and diagnosis applications are another focus. Traffic applications, including vehicle assistance systems are significant to manufacturers with 25% and 13%, but are less relevant to the user with 10% and 2%.
Within industry sectors, mechanical engineering and the automotive industry as well as electronics and measurement technology dominate areas of use. Research and development, medical technology and the health sector follow. Other noteworthy areas of use are the food industry, printing and packaging, security and surveillance as well as the general area of logistics.
Independent implementation supports willingness to invest in Embedded Vision
Compared with last year, manufacturers are selling significantly more to direct customers, 40% this year, compared with 24% in 2015. From this, it can be deduced that increased user-friendliness in implementation and operation is intertwining perfectly with the required individualization. The meshing doesn’t necessarily play into the hands of using smart cameras, in the coming 2 years, users are expecting a small increase from 27% to 34%. The increase according to manufacturers remains low, from 13% to 20%. This is not a surprise because manufacturers sell 30% to system integrators and OEMs who have their own technical know-how and application-specific specialist requirements. Compared with last year, the use of preconfigured smart cameras stagnated at 27%. For simple checking tasks and processes with low complexity, however, these preconfigured compact systems will probably remain popular components that are easy to integrate. With increased integration of image processing systems into the PLC of very individual systems and complex manufacturing processes, also known as “Embedded Vision“, the use of modular image processing systems with a central control unit is becoming necessary. The more cameras are being used with increasing numbers of tasks and increased complexity of the overall application, the more it makes sense (becoming more cost-effective) to use centralized controllers that analyze several cameras and dependent systems across different tasks.
Just like in previous years, willingness to invest in image processing is still high. Driven by digitalization and automation in the factory of the future and the use of visual measuring technology in science, surveillance and entertainment, 100% of all manufacturers and 90% of all users stated that they would want to introduce new image processing based systems or modernize existing systems over the coming 2 years. 64% of all users want to develop and implement their own systems. With a view to complete linking of production and processes, thinking intelligent systems are most important to users where embedding is uncomplicated, yet allows a high degree of individualization. The use of image-based software intelligence, so-called Deep Learning, is still in its infancy but it represents most potential for strategic use in image processing technology.
Stable price range and high level of brand recognition
Whether the market will grow reliably, depends on the price range, in addition to general willingness to buy. After a high of 70% in 2014, manufacturers only want to produce about 44% of cameras in the mid-price range between $150 and $650. This was also the case in 2015. This shows stabilization after successive price drops in the previous years. Cheap cameras under $150 are least significant to manufacturers and users in terms of percentage, at 26% and 11%, respectively. Compared with 2015, high-priced cameras from $650/$1000/$3000 have dropped by 12% points. This price drop refers exclusively to the upper mid-price range between $650 and $1000, strengthening the mid-price range owing to a shift to cameras from $350 based on an increased technological standard. These days, cheaper cameras suffice for tasks of average complexity. The stable proportion of cameras from $1000/£3000 for manufacturers (19%) and users (23%) shows that high quality and individualized cameras are still required for complex networked automation processes and embedded optical control systems.
Among users, the best known camera brand is e2v with a supported brand recognition of 52%. It can be assumed, however, that a percentage of this distinction comes from the sensor sector. Matrox, Teledyne Dalsa and Cognex follow densely, each have 46%. Sony cameras are used most (31%); Basler follows with 27%, and Teledyne Dalsa and IDS with 25% each.
Change to CMOS is the reason for increase in custom sensors – Sony and On Semi remain strong
Based on the discontinuation of CCD sensors by Sony within the next 5-10 years, about 40% of manufacturers and users want to switch to CMOS technology, similar to last year. This no longer concerns almost 2/3 of manufacturers and 1/3 of users because they almost exclusively rely on CMOS sensors. The strategic decision by Sony seems to have only had an effect on manufacturers at this time. As opposed to the almost undetectable drop from 42% to 40% in 2015, 32% of all camera manufacturers rely on Sony these days, but, in 2 years, want to get back to the level before the discontinuation, with 37%. When it comes to users, even more rely on Sony compared to last year (35% increased to 53%), and the forecast drop to 25% can be ignored completely. 51% want to remain loyal to the market leader for the next two years as well. This means that manufacturers as well as users are showing trust in Sony and its technological strategy as well as model development. The majority of users and manufacturers state that they have not been affected by delivery delays caused by the earthquake in Japan. Only 13% of manufacturers are experiencing severe delays. Based on the CCD shift, e2V benefits with reference to manufacturers (from 3% to 12%) and customer-specific sensors (from 4% to 19%) with an equally positive forecast. Compared to last year, On Semiconductor, second in the market, managed to stay stable in terms of manufacturers with only a small drop (34% to 29%) and a moderate growth in users (from 11% to 16%). The company will remain at the current level, according to forecasts.
Manufacturers are selling as much as 85% of cameras with CMOS technology. Users want to reach this purchasing level in the next 2 years. With no increase compared to last year, 51% of users rely on CMOS today but are predicting a faster growth to 83% (as opposed to 70% in 2015). It is possible that many companies have already started with the planning and designing phases, but the Go-live with new sensors and systems may still lie ahead of them.
For embedded systems: Small sensors with high resolution and matching lenses
Along with miniaturization, the technological development of CMOS sensors with smaller pixel sizes, higher resolution and improved sensitivity promotes the use of image processing in integrated systems. Augmented Reality based data glasses for effective production maintenance and assistance systems for human-machine collaboration in robotics as well as autonomous transport systems in the areas of traffic, automotive and logistics, are examples of high quality, miniaturized and embedded image processing intelligence for the control and surveillance of the digital economy. As predicted, a shift to the next level of megapixel resolution has occurred. In 2015, around 30% of users relied on sensors under 1 MP. In 2016, it’s only 10%. The loss in the smallest category can be marked as growth in the categories between 1 and 3 MP (+10% points) as well as in the categories between 3 and 5 MP (+2% points) and between 5 and 10 Megapixels (+3% points). In 2 years, users are expecting a wide focus in the area between 5 and 10 Megapixels for sensors, with an increase from 11% to 26%. Camera manufacturers are planning more carefully, with an expected growth between 1 and 5 Megapixels and a continuing significant and stable proportion below 1 Megapixel. The most popular read out technology is the distortion-free global shutter method, with 55% of manufacturers and 77% of users.
Users and manufacturers do agree on sensor miniaturization: The category of sensors between 1/3″ and 1/2″ increased from 20% to 49% for users this year. Manufacturers are also reaching the highest value of their sales in this size category with 41% points. Their focus is clearly in the area of miniaturization with an additional 17% of sensors under a size of 1/3″, likely supported by the OEM business with Embedded Systems. To complement this, manufacturers are selling more than half of their cameras with CS-Mount or mini lenses, while users corresponded with almost a third. C-Mount lenses currently remain the most used lenses with 41% according to manufacturers and 59% according to users.
As one of the most important quality criteria within an image processing system, the awareness and use of lenses was surveyed for the first time in the 2016 FRAMOS Market Survey. Unsurprisingly, Zeiss lenses are known to most users, at 56%, followed by Tamron and Fujinon at 44% each. However, Fujinon lenses lead in use, with 44%, followed by Kowa (29%) and Schneider-Kreuznach (27%).
Interfaces for increased data volumes and real-time processing
The main advantage of Embedded Vision Systems is real time surveillance and analysis. Fast frame rates and their rapid transmission for immediate processing are required for high precision. Manufacturers as well as users expect a focus on frame rates between 25fps and 60fps today and in the coming two years. At the same time, compared to last year, significant increases in the area of over 100fps (+13% points for users) and 200fps (+14% points for manufacturers) have taken place.
In today’s interface use of GigE Vision dominates according to manufacturers, with 33%, followed by Ethernet with 15%. Compared with last year and based on a low proportion of American participants, the previously high Ethernet percentage has been reduced. Manufacturers as well as users are expecting most growth for USB 3.0 and GigE Vision with an increase of 8% and 10% points, respectively. More than 75% of manufacturers and, with 60% a majority of users are expecting that bandwidths of more than 5GB will become relevant or very relevant for image processing applications. 50% of manufacturers believe that USB 3.1 is the most important interface for High Speed applications, followed by 38% for 10 GigE. Contrary to this, 44% of users favor 10 GigE for fast transmission, whereas 37% of users vote for USB 3.1. Today, 35% of all users utilize frame grabbers to process these frame rates and bandwidths as well as connecting the interfaces of the camera and the processor unit, while expecting an equal demand within the next 2 years.
Robust, intelligent systems increase scalability and profitability
“The most interesting part of the Market Survey each year, in addition to the technical information, is the text boxes in which participants can note which factors they deem necessary for further growth and where the biggest potential lies,” says Dr. Andreas Franz, patron of the survey. “The anonymized data provides a very detailed picture of what manufacturers and users feel about the market situation and requirements.”
Simple system integration and uncomplicated use for non-programmers is crucial for growth and the increased utilization of image processing technology by users. As most users implement their systems themselves, they would like thinking automated Vision Systems that can seamlessly be embedded in their overall systems and can be individually adapted. They place particular importance on industry standards, especially in the area of interfaces, as well as a high level of data security within networked systems. Since image processing has arrived as the eye of automation in the mid-sized sector, SMEs, in particular, are hoping for the intelligent use of improved algorithms and libraries for profitable scalability of their systems, preferably as open source offers. Image processing should no longer exclusively optimize the production flow but has to contribute as an analysis tool to process optimization and increased ROI.
According to users, the foundation will remain consistent with technical development of sensors and cameras in the area of resolution, speed and miniaturization, with decreasing prices. However, the demand for increasingly improving quality characteristics for more precise evaluations is much stronger than for “cheaper cameras”. Based on the virtually exhaustive application options of visual sensor systems in all industries as well as research and surveillance, the robustness of the technology is hugely important. When used outdoors, wide temperature ranges are required, and Industry 4.0 and heavy industry demand reliability in the event of pressure changes or vibration. Medical applications as well as aerospace require high radiation resistance of the components used.
Augmented Reality and Deep Learning as strategic areas of growth
For the first time since the beginning of the Market Survey in 2007, users see a specific relevance for the practical use of Augmented Reality applications in the area of User Guidance and digital support offers. Huge potential is mainly awarded to so-called “assistance systems” in industrial maintenance as well as autonomous traffic and logistics systems. Manufacturers also agree with the issues of networked automation, robotics and human-machine collaboration. Networking multiple camera units across several systems in the Cloud as well as the associated data use for comprehensive Deep Learning analysis is also mentioned for the first time as a strategic area of growth for image processing technology and its intelligent use. Users are hoping for support from the thinking system itself, simple configurability and independent cognitive learning processes of their current systems.
The results of the 2016 FRAMOS Market Survey show that image processing managed to leave the niche market a long time ago. Users and manufacturers recognize comprehensive areas of application for this technology across the entire digital supply chain in industry, science, and security. Visual sensor systems, as the intelligent eye of automation, let images talk and machines think. In the area of thinking and controlling, in particular, based on Deep Learning we can now begin to sense the immense potential image processing has as integrated technology and strategic-analytical decision support for increased price effectiveness and efficiency.