A machine vision camera not only needs to capture a sharp image, but above all an image at the right moment. With a stationary product, this is usually straightforward. As soon as products move, conveyor speed varies or lighting needs to flash briefly, timing becomes an important part of the system design.
An incorrectly chosen trigger strategy does not usually create a classic software problem, but an acquisition problem: the product is not reproducibly positioned in the same place in the image. That is why triggering should be considered early in the system setup.
Why triggering is important in machine vision
The choice of the right trigger is closely linked to the camera, lighting, exposure time, interface, software and the mechanics around the product. Anyone building a vision system must therefore determine not only which camera provides sufficient image quality, but also when that camera should capture the image and how that moment can be kept reproducible.
Free-running image acquisition
With free-running image acquisition, the camera continuously captures images as soon as acquisition has started. After an image has been captured and read out, the next acquisition automatically follows. No external sensor or trigger pulse is required to start each individual image.
This mode is particularly suitable when the product is stationary or when the exact acquisition moment is not critical. Examples include a test setup, a manual inspection position or a component that is mechanically positioned before inspection starts. The vision software then works with the images made available by the camera.
The maximum frame rate stated on the datasheet applies to a specific camera setting and readout mode. In practice, this value must always be assessed together with sensor readout, interface bandwidth and exposure time. If the exposure time or transfer time becomes longer than the available time per image, the effective image rate decreases.
The advantage of free-running acquisition is simplicity. With a GigE camera with PoE, a single network cable can be sufficient for power and data, provided that the camera, cable and network port support PoE. With a USB3 camera, a single USB cable may be sufficient. This simplicity is attractive, but only as long as the inspection moment does not need to be accurately linked to product movement.
Triggered image acquisition
With triggered image acquisition, the camera only starts after a trigger event has been received. This event can come from the software or be supplied via a physical input on the camera. The camera then starts the acquisition, optionally with a configured trigger delay.
The advantage is that image acquisition is linked to an event in the machine. A product passes a photoelectric sensor, reaches a fixed position or is released by the control system for inspection. The camera does not simply capture images continuously, but records when the product is in the desired position.
This is particularly important for moving products. When a product moves through the image at speed, even a small timing variation determines whether the relevant feature appears in the same position. The software can sometimes correct positional differences, but an image captured too early, too late or with motion blur remains a weak basis for inspection.
Software trigger or hardware trigger
With a software trigger, the vision software sends a command to the camera to start an acquisition. This is practical for test setups, laboratory systems or slow-moving processes. The setup remains simpler because no additional trigger wiring is required.
With hardware triggering, a sensor, often a photoelectric sensor, detects that a product is present. The signal is sent to a digital input on the camera. The camera then starts the acquisition. This method is widely used in production environments because the acquisition moment is linked more directly to the machine.
The position of the photoelectric sensor is important. If the sensor is too close to the camera, there is little time between detection and acquisition. If it is too far away, accuracy becomes more sensitive to speed variation or slip on the conveyor. At constant speed, a fixed trigger delay may be sufficient. At variable speed, a fixed delay alone is often less reliable.
In surface inspection, it is also important to prevent the photoelectric sensor itself from causing interference in the image. The light spot or reflection from the sensor may become visible on glossy products, film or packaging. In practice, this can be solved by repositioning the sensor, triggering on a different signal edge or positioning the photoelectric sensor so that its light does not enter the camera image.
Encoder triggering for line scan cameras and continuous material
With line scan cameras, an encoder is often used as a line trigger. The camera then does not capture a new line based on time, but based on the actual movement of the conveyor, roller or material web. Each encoder pulse, or a configured division of those pulses, determines when the next line is acquired.
This is especially important when inspecting continuous material such as film, paper, textile, metal strip or other continuous surfaces. If the line trigger is time-based and the speed varies, the distance between the acquired lines changes. The image is then stretched or compressed in the direction of movement. With an encoder, line acquisition remains linked to the actual displacement of the material.
With continuous material, the frame start is often initiated by software when the inspection needs to become active. The line trigger then comes from the encoder. The camera continues to acquire lines as long as acquisition is active and encoder pulses are supplied.
For individual products scanned with a line scan camera, the frame start may instead come from a hardware trigger, for example from a photoelectric sensor that detects that the product is approaching the scan position. The encoder still provides the line trigger during product movement.
A line scan camera therefore often has two timing levels. The frame start determines when the scan begins. The line trigger determines the spatial step with which the lines are acquired. If these two signals are not selected correctly, the product may appear in the image, but it will not be built up geometrically correctly.
Multiple images after one trigger
Some applications require not one image, but multiple images after a single trigger. The camera receives one start signal and then captures a sequence of images. This can be useful when a product passes multiple positions, when an object moves or rotates, or when several images are required for one evaluation.
This mode requires attention to timing, data transfer and software logic. The software must know which image belongs to which position. If different lighting setups are used, the lighting control must also be synchronized with the image sequence. Otherwise, the correct product may be evaluated with the wrong lighting.
Triggering as part of the system design
Triggering cannot be selected separately from the rest of the vision system. A shorter exposure time reduces motion blur, but usually requires more light. More light may mean that controlled or strobed lighting is required. A higher frame rate requires more interface bandwidth and processing capacity. A higher resolution provides more detail, but also increases the amount of data per image.
With GigE and USB3 cameras, the trigger is often connected directly to the camera. The camera captures the image and then sends it to the PC or vision processor. With Camera Link or CoaXPress systems, the trigger may also be part of the frame grabber or acquisition configuration. The most logical solution depends on the camera interface, speed, synchronization requirements and software environment.
Anyone building a vision system must assess these choices together. Camera, lighting, software, mechanics and timing jointly determine whether the image is reproducible enough for inspection. More information about this approach to system design can be found on the page about integrating a machine vision system yourself.
Conclusion
Free-running image acquisition is simple and suitable for stationary products or situations where timing is not very critical. As soon as products move or the acquisition moment must be accurately linked to the product position, triggering becomes an essential part of the system.
Software triggering is useful for testing and less critical applications. Hardware triggering is widely used in production environments because the acquisition moment is linked more directly to the machine. With line scan cameras and continuous material, the encoder becomes important as a line trigger because acquisition is then based on actual displacement rather than time.
A trigger is therefore not a setting to be chosen only at the end of the project. It determines whether the image is captured at the right moment, with the right lighting and at the correct product position. Reliable inspection starts not only with a good camera, but with an acquisition setup that technically matches the movement of the product and the timing of the machine.
