Camera Technologies in Machine Vision

Sensor type: CMOS vs. CCD

The type of sensor plays a central role in determining both the image quality and the applicability of the camera. Although CCD sensors were traditionally used in machine vision, CMOS sensors have become the standard for most industrial applications thanks to their higher sensitivity, lower cost and reduced power consumption. CMOS sensors also enable higher frame rates and are better suited to integration into compact systems.

Resolution defines the smallest detectable feature.

Frame rate impacts inspection speed and timing.

The frame rate, measured in frames per second (fps), affects the speed and efficiency of the inspection. In applications involving the quick movement or processing of objects, a high frame rate is crucial to ensure that all relevant information is captured without any image data being lost. An inadequate frame rate can lead to processing delays and inaccurate results.

Shutter type: global for moving parts and rolling for static inspections.

Interface: USB3, GigE Vision, CoaXPress or Camera Link.

The camera interface influences not only the data transfer speed, but also compatibility with existing systems. USB3 is ideal for simple applications with low bandwidth requirements, while GigE Vision and CoaXPress are better suited to more demanding applications requiring higher data rates and longer cable lengths.

Camera Link offers high bandwidth and is commonly found in industrial applications where real-time processing is required. The choice of interface should be based on the application's specific requirements.

Housing: compact, IP-rated or temperature-controlled for harsh environments.

• Camera selection and specification
• Interface integration and cabling
• Mechanical mounting and optical alignment
• Multi-camera synchronization for complex views

For space-constrained or high-volume applications, compact OEM camera modules are available. These allow deeper integration in robotics, medical devices, and consumer equipment. STEMMER IMAGING supports hardware customization, thermal management and regulatory considerations.

• Performance calculations, interface matching, camera testing

• Pre-mounted camera modules, aligned optics, calibration

• Custom housings, embedded camera designs, cleanroom-ready integration

The most common types are area scan cameras and line scan cameras. Area scan cameras capture a full rectangular image in a single exposure and are the standard choice for inspecting discrete objects or scenes. Line scan cameras capture one line at a time, building up an image as the object moves past - they are essential for continuous materials such as webs, films, textiles, or cylindrical surfaces. Beyond these, 3D cameras, smart cameras, and thermal cameras serve specific application requirements.

An area scan camera captures a two-dimensional image in a single frame. It is well suited for static or moving objects that can be imaged in a snapshot. A line scan camera uses a single row of pixels to capture images progressively as the object passes the camera. This makes line scan ideal for high-resolution inspection of continuous or rotating objects, where an area scan camera would require multiple, overlapping images to achieve the same coverage. Line scan also delivers very high resolutions without the cost and data overhead of a large area sensor.

The right interface depends on bandwidth requirements, cable length, and whether a frame grabber is needed. USB3 Vision is a practical choice for low-to-medium bandwidth applications with short cable runs. GigE Vision is the most widely used standard, offering Ethernet distances of up to 100 metres without a frame grabber. For very high-speed applications requiring maximum data throughput, CoaXPress or CameraLink provide the necessary bandwidth but require a compatible frame grabber card. All current standards sit under the GenICam umbrella, ensuring consistent camera control across platforms.

Resolution requirements are determined by the smallest feature that must be reliably detected, the size of the field of view, and the required inspection accuracy. A common starting point is to ensure the feature occupies at least 2–3 pixels in the image. For example, detecting a 0.1 mm defect on a 100 mm wide part requires a camera with at least 1,000 pixels across that dimension. Resolution must always be balanced against frame rate, sensor size, and available lighting - increasing resolution without adjusting other parameters typically reduces frame rate.

GigE Vision is a machine vision standard that defines how cameras transmit image data over standard Gigabit Ethernet infrastructure. Its widespread adoption is driven by the availability of Ethernet hardware, cable lengths of up to 100 metres without a frame grabber, and the ability to run multiple cameras on a single network. GigE Vision is part of the GenICam framework, which standardises camera configuration across different manufacturers. Variants including 2.5GigE, 5GigE, and 10GigE address applications requiring higher bandwidth.

Yes. Industrial cameras are designed for demanding environments. IP-rated housings protect against dust and water ingress, and many models are available in compact, vibration-resistant configurations. For applications with extreme temperature ranges, pressure, or chemical exposure, dedicated protective enclosures can house standard cameras. When selecting a camera for a harsh environment, it is important to consider both the camera's IP rating and its operating temperature range, and to plan cable routing and connector protection accordingly.