No-one should expect the enabling qualities of machine vision to come free of charge but Jason Barnes finds there is still much that ITS stakeholders can do to help reduce costs. After many years of application in high-end solutions for the enforcement and tolling sectors, machine vision is gaining traction in more general areas of traffic management. Nevertheless, those OEMs producing transport-oriented solutions which incorporate machine vision and looking to increase the technology’s share of the ITS mar
The proliferation of the more sophisticated vision-based solutions for traffic management would be greatly facilitated by a more collaborative approach to standards
No-one should expect the enabling qualities of machine vision to come free of charge but Jason Barnes finds there is still much that ITS stakeholders can do to help reduce costs.
After many years of application in high-end solutions for the enforcement and tolling sectors, machine vision is gaining traction in more general areas of traffic management. Nevertheless, those OEMs producing transport-oriented solutions which incorporate machine vision and looking to increase the technology’s share of the ITS market face some enduring challenges. The multiplicity and sophistication of potential applications is leading to an increasingly diverse range of systems and subsystems which can be incorporated into a total solution. This increases the need for all those in the machine vision value chain – OEMs, tier suppliers (including camera and peripherals providers) and end users – to work together more closely to optimise products. It is also leading to evermore sophisticated and subtle interplays between the various elements within a solution. The statement that there is no one wholly right way of doing things has never been truer.
There is also a keenness among machine vision camera manufacturers to stress the availability of products within their ranges which are intended to handle the highly variable light conditions which prevail with outdoor applications such as traffic management. This is by comparison with, for example, manufacturing processing within a tightly controlled, indoor environment.
There has also been significant resource pushed into reducing the cost point of entry to the technology. Machine vision manufacturers recognise that while their traditional markets accept a given level of performance comes at a given price, this is harder to sell to the ITS and security markets which until recently saw relatively unsophisticated, analogue CCTV solutions as their performance, and price, benchmark. Where CCD-based camera solutions were seen as the only realistic choice just a few years ago, there is now healthy competition from CMOS sensor-based alternatives which offer higher speed performance, better integration potential, need less power and enjoy faster development cycles. CMOS solutions’ single-chip nature has had a downward pressure on price, as has their roots in consumer products, which brings volume production into the picture.
For the sake of balance, it is worth highlighting that CCD-based cameras remain in favour for certain applications. Whatever the core technology, manufacturers continue to emphasise the need for commercial-grade solutions for demanding applications such as transport management; resilience may demand a certain premium but when consumer-grade systems are compared in a whole-life context commercial-grade remains a competitive option.
Concurrently, as machine vision technology branches out into new application areas, a range of connectivity standards are being developed and adopted by the wider market. Those such as Camera Link can be considered mature and perhaps even of limited currency as newer standards such as GigE Vision, USB 3 Vision, Precision Time Control and others continue to gain ground. Collectively, these standards give designers significant latitude in terms of network layout, allowing trade-offs to be made in terms of data transmission distances, camera performance and so on. This too has positive cost implications, in that a series of lower-cost (but still relatively capable) cameras can be used to substitute for a single, or very small number, of very high-performance models.
The proliferation of applications and of the solutions intended to serve them makes increasing specialisation inevitable. Lighting provides an example. As even the most inexperienced of photographers can attest, the better the lighting conditions, the less capable a camera need be. The same holds true for the rather more sophisticated machine vision offerings.
Sourcing from specialist lighting manufacturers is not always considered by OEMs – who think nothing of sourcing from specialist camera manufacturers – yet lighting remains highly relevant where the aim is to optimise overall solution performance and minimise cost. This example highlights how a more collaborative approach is needed, both within the machine vision sector itself and between the machine vision sector and those within newer areas of application.
At the same time, the emergent sectors of application need to think about how they will use machine vision. The situation has been reached where the machine vision industry needs to see the ITS sector heading back towards it with a specific set of standards which define how it wants to use these sophisticated, vision-based technologies.
This is not just a case of developing standards for standards’ sake. Take the work done within European Framework Projects to formulate and implement, for example, the DATEX II standard for information exchange between traffic management centres, or the analogous creation and deployment of the NTCIP communications standard in the US. Both show how common standards facilitate greater ease of operations and, along the way, help to build volume of production, expand a common knowledge base and – perhaps most crucially – reduce costs.
Lighting: defining the need and the solutions
The subtle interplays between camera, lens assembly and lighting all have to be considered within the context of a specific application. However, when artificial lighting is used, the system designer is concerned with how the OEM/user wants light delivered to the object to be monitored and with achieving very high levels of uniformity, stability and repeatability and, increasingly, light intensity.
An answer is to hand responsibility to an outside technology Provider with expertise in LED lighting and lighting control encompassing products, standards and the ability to guarantee component supply, says6965 Gardasoft Vision’s Peter Bhagat.
“Maintaining technical currency and producing leading-edge lighting solutions requires considerable investment in what may only be a single element of a very complex, vision-based end product. The result can be to divert from core competencies.
The interplays are complex. For example beam angles are a product of power output and optics, lenses being matched pairs with particular LEDs. “Simply arranging a large number of small-chip LEDs on a PCB may give very good light density but a high-brightness LED solution may be more efficient if the right optics are used.”
“Also, non-specialists can be starting from scratch. That can mean beginning with iteration one when specialists are perhaps already a dozen years and several iterations on. Moving from iteration one to iteration two may only bring slight changes in capability when a step change is required.”
But, he says, he already sees signs of the cooperation needed to take machine vision into wider application.
“Gardasoft recently provided an InfraRed (IR)-based lighting solution designed to help penetrate IR-reflective windscreens in order to detect printed licences. Such licences can be susceptible to fraudulent copying and there was a need to be able to ascertain whether the article displayed is in fact genuine. The solution arrived at could only have come about as a result of close cooperation between the end user, OEM and tier supplier. Partnerships are emerging and are already being successful.”
After many years of application in high-end solutions for the enforcement and tolling sectors, machine vision is gaining traction in more general areas of traffic management. Nevertheless, those OEMs producing transport-oriented solutions which incorporate machine vision and looking to increase the technology’s share of the ITS market face some enduring challenges. The multiplicity and sophistication of potential applications is leading to an increasingly diverse range of systems and subsystems which can be incorporated into a total solution. This increases the need for all those in the machine vision value chain – OEMs, tier suppliers (including camera and peripherals providers) and end users – to work together more closely to optimise products. It is also leading to evermore sophisticated and subtle interplays between the various elements within a solution. The statement that there is no one wholly right way of doing things has never been truer.
Moves towards ITS
That the machine vision industry is serious about making inroads into ITS is beyond doubt. There was a noticeably increased presence of machine vision system manufacturers exhibiting in their own right – as distinct from having their products buried deep within a traffic management system OEM’s end solution – at this year’s Intertraffic, for example, and non-traditional machine vision applications have enjoyed greater emphasis at recent machine vision-oriented events such as the Vision show. This is a clear demonstration of how the machine vision sector views the traffic management/ITS sector as having significant potential for future growth.There is also a keenness among machine vision camera manufacturers to stress the availability of products within their ranges which are intended to handle the highly variable light conditions which prevail with outdoor applications such as traffic management. This is by comparison with, for example, manufacturing processing within a tightly controlled, indoor environment.
There has also been significant resource pushed into reducing the cost point of entry to the technology. Machine vision manufacturers recognise that while their traditional markets accept a given level of performance comes at a given price, this is harder to sell to the ITS and security markets which until recently saw relatively unsophisticated, analogue CCTV solutions as their performance, and price, benchmark. Where CCD-based camera solutions were seen as the only realistic choice just a few years ago, there is now healthy competition from CMOS sensor-based alternatives which offer higher speed performance, better integration potential, need less power and enjoy faster development cycles. CMOS solutions’ single-chip nature has had a downward pressure on price, as has their roots in consumer products, which brings volume production into the picture.
For the sake of balance, it is worth highlighting that CCD-based cameras remain in favour for certain applications. Whatever the core technology, manufacturers continue to emphasise the need for commercial-grade solutions for demanding applications such as transport management; resilience may demand a certain premium but when consumer-grade systems are compared in a whole-life context commercial-grade remains a competitive option.
Concurrently, as machine vision technology branches out into new application areas, a range of connectivity standards are being developed and adopted by the wider market. Those such as Camera Link can be considered mature and perhaps even of limited currency as newer standards such as GigE Vision, USB 3 Vision, Precision Time Control and others continue to gain ground. Collectively, these standards give designers significant latitude in terms of network layout, allowing trade-offs to be made in terms of data transmission distances, camera performance and so on. This too has positive cost implications, in that a series of lower-cost (but still relatively capable) cameras can be used to substitute for a single, or very small number, of very high-performance models.
Interplays with specialism
Many positive developments can be defined but that is not to say the journey is done. This enabling technology grew out of a highly specialised market. Even within the machine vision sector itself it remains the domain of multiple experts – and what holds true for traditional applications is perhaps doubly so for newer application sectors such as ITS.The proliferation of applications and of the solutions intended to serve them makes increasing specialisation inevitable. Lighting provides an example. As even the most inexperienced of photographers can attest, the better the lighting conditions, the less capable a camera need be. The same holds true for the rather more sophisticated machine vision offerings.
Sourcing from specialist lighting manufacturers is not always considered by OEMs – who think nothing of sourcing from specialist camera manufacturers – yet lighting remains highly relevant where the aim is to optimise overall solution performance and minimise cost. This example highlights how a more collaborative approach is needed, both within the machine vision sector itself and between the machine vision sector and those within newer areas of application.
Broader perspectives
Within the machine vision sector, there needs to be an even greater concentration on modularity and the much-quoted ‘interconnectedness of things’. As the technology continues to move away from discrete solutions and applications, it needs to be able to work with or alongside solutions from a wide range of suppliers in an increasing number of different settings. This is an issue which is far wider than just the machine vision and ITS sectors; there is little reason why common standards cannot be developed which will allow openness across many applications and bring about true economies of scale.At the same time, the emergent sectors of application need to think about how they will use machine vision. The situation has been reached where the machine vision industry needs to see the ITS sector heading back towards it with a specific set of standards which define how it wants to use these sophisticated, vision-based technologies.
This is not just a case of developing standards for standards’ sake. Take the work done within European Framework Projects to formulate and implement, for example, the DATEX II standard for information exchange between traffic management centres, or the analogous creation and deployment of the NTCIP communications standard in the US. Both show how common standards facilitate greater ease of operations and, along the way, help to build volume of production, expand a common knowledge base and – perhaps most crucially – reduce costs.
Lighting: defining the need and the solutions
The subtle interplays between camera, lens assembly and lighting all have to be considered within the context of a specific application. However, when artificial lighting is used, the system designer is concerned with how the OEM/user wants light delivered to the object to be monitored and with achieving very high levels of uniformity, stability and repeatability and, increasingly, light intensity.
An answer is to hand responsibility to an outside technology Provider with expertise in LED lighting and lighting control encompassing products, standards and the ability to guarantee component supply, says
“Maintaining technical currency and producing leading-edge lighting solutions requires considerable investment in what may only be a single element of a very complex, vision-based end product. The result can be to divert from core competencies.
The interplays are complex. For example beam angles are a product of power output and optics, lenses being matched pairs with particular LEDs. “Simply arranging a large number of small-chip LEDs on a PCB may give very good light density but a high-brightness LED solution may be more efficient if the right optics are used.”
“Also, non-specialists can be starting from scratch. That can mean beginning with iteration one when specialists are perhaps already a dozen years and several iterations on. Moving from iteration one to iteration two may only bring slight changes in capability when a step change is required.”
But, he says, he already sees signs of the cooperation needed to take machine vision into wider application.
“Gardasoft recently provided an InfraRed (IR)-based lighting solution designed to help penetrate IR-reflective windscreens in order to detect printed licences. Such licences can be susceptible to fraudulent copying and there was a need to be able to ascertain whether the article displayed is in fact genuine. The solution arrived at could only have come about as a result of close cooperation between the end user, OEM and tier supplier. Partnerships are emerging and are already being successful.”
