IMAGE SENSOR TECHNOLOGY / MACHINE VISION
A camera can deliver sharp images of stationary objects and still produce inaccurate shapes when the object or camera moves. In many machine vision systems, the problem is not resolution, focus or USB bandwidth. It is the way the image sensor exposes different parts of the scene.
Choosing between a global shutter camera and a rolling shutter camera affects motion distortion, measurement reliability, exposure control, lighting synchronization and the image sensor options available for a project.
This article compares the two shutter architectures from an engineering perspective, with practical examples for conveyor inspection, robotics, barcode reading, automated measurement and embedded vision. It also explains why global shutter does not automatically eliminate motion blur.
CHOOSE GLOBAL SHUTTER WHEN
The object or camera moves during image capture and the application requires consistent object geometry, accurate positioning or precisely timed image acquisition.
CHOOSE ROLLING SHUTTER WHEN
The scene is stationary or moves slowly, geometric distortion is acceptable or controllable, and sensor resolution, image quality, cost or availability is the stronger priority.
Important: Neither shutter type is automatically better in every application. Exposure time, sensor readout timing, illumination, lens performance and the host system must all be evaluated.
The main difference is when different image sensor rows collect light.
In a global-shutter sensor, the pixels capture the scene during the same exposure interval. The image data can then be read out sequentially without introducing the row-by-row exposure-time offset characteristic of rolling shutter.
In a rolling-shutter sensor, different rows begin and end exposure at different times. When the scene changes during that interval, the final image combines information captured at different moments.
The diagrams below are conceptual illustrations. Real sensor timing depends on the selected sensor and readout mode.
All rows record light over the same interval.
Each row captures a different interval of the changing scene.
A critical distinction is that simultaneous exposure does not necessarily mean simultaneous data readout. Many global-shutter CMOS sensors capture the complete frame at once and then read pixel data sequentially.
Therefore, shutter architecture should not be confused with total sensor readout time, maximum FPS or end-to-end camera latency.
Rolling shutter artifacts appear when the scene changes between the exposure of different sensor rows. A moving object may have a different position in the upper part of the image than in the lower part.
The effect depends on object speed, camera motion, scan direction and the time offset between rows. It can appear even when each individual row is exposed briefly.
Straight vertical objects appear tilted when the camera or object moves horizontally during row-by-row capture.
Rapid vibration or changing camera orientation can make straight structures appear curved or unstable between image rows.
Short flashes or rapidly modulated light can affect different rows differently, creating horizontal bands or uneven brightness.
In ordinary video, a slightly tilted object may be visually acceptable. In a measurement or robotic positioning system, however, geometric distortion can affect the location and shape estimated by the vision algorithm.
Consider a moving rectangular component on a conveyor. If the image contains different positions of the component at different row times, the resulting rectangle may appear skewed. A vision algorithm measuring its edges may then report an inaccurate position or shape.
Shortening exposure time can reduce motion blur within each row, but it does not by itself remove the time difference between rolling-shutter rows. To reduce rolling-shutter geometric distortion, engineers may need faster sensor readout, controlled motion, synchronized illumination or a global-shutter sensor.
No. A global-shutter sensor can still produce motion blur.
Global shutter prevents the sequential-row exposure timing from distorting the geometry of a moving subject. But during the common exposure window, the object may continue to move. If the exposure is too long, its image will still be blurred.
Motion blur should therefore be evaluated separately from rolling-shutter distortion.
| Imaging Problem | Main Cause | Primary Solution Direction |
|---|---|---|
| Motion Blur | Object or camera movement during exposure | Shorter exposure, more suitable illumination and optical configuration |
| Rolling-Shutter Skew | Different sensor rows capture at different times | Faster rolling readout, synchronized lighting or global shutter |
| Low-Light Noise | Insufficient signal, electronic noise and gain | Lighting, sensor sensitivity, exposure and appropriate image processing |
Suppose an object moves along a conveyor at 1 meter per second. Its travel distance during one exposure can be approximated by:
Object Travel During Exposure = Object Speed x Exposure Time
At 1ms exposure
1mm
Object-plane travel during the exposure interval
At 100 microseconds
0.1mm
Object-plane travel during the exposure interval
These numbers describe physical movement in the object plane, not a guaranteed blur width in image pixels. Pixel-level blur depends on the imaging magnification and effective object-space pixel size.
If the inspection tolerance requires very limited movement during capture, shorter exposure may be necessary regardless of whether the sensor is global or rolling shutter.
When shorter exposure reduces image brightness, controlled illumination, lens aperture and sensor sensitivity become part of the shutter-selection decision.
Two rolling-shutter sensors operating at the same 30fps can produce different levels of motion distortion. The reason is that the time between exposure of the first and last relevant rows may be different.
A fast-reading rolling-shutter sensor can capture a moving scene with less geometric distortion than a slower-reading rolling-shutter sensor, even if both report the same output frame rate.
Assume a rolling-shutter configuration has a time offset of 20 microseconds between successive row exposure starts.
1,000 relevant rows x 20 microseconds
Approximately 20ms
This is an illustrative first-to-last-row timing span, not a measured value for a CK Vision sensor. The exact span between 1,000 row starts would use 999 intervals. Actual timing depends on the sensor mode, selected region of interest and readout architecture.
During this interval, a fast object or moving camera can change position significantly. The resulting geometric distortion depends on the direction and magnitude of that motion.
When comparing rolling-shutter sensors, ask for the applicable row time, effective readout timing and operating mode rather than using maximum FPS as the only motion-performance metric.
Shutter type, triggering and lighting control are related, but they are not interchangeable functions.
A global-shutter sensor captures all pixels within the same exposure interval. A hardware trigger determines when an acquisition or exposure event is initiated, according to the camera's supported trigger mode. A strobe provides controlled illumination during a defined time interval.
Controls the temporal relationship between pixel exposures across the sensor.
Controls the timing of image acquisition when the selected camera supports external triggering.
Controls when sufficient light reaches the subject during the intended exposure window.
A position sensor detects a component approaching the inspection point. A control system sends a trigger to a compatible camera, and a synchronized light illuminates the target during image exposure.
This type of system benefits from predictable acquisition timing. However, selecting a global-shutter sensor does not automatically provide a hardware trigger input, GPIO, strobe output or deterministic host latency. These functions must be supported by the complete camera design.
In some camera systems, yes. A suitable rolling-shutter sensor and lighting controller can use a brief flash during a period when all relevant rows are simultaneously integrating light. With carefully controlled ambient illumination, this can reduce motion-related distortion.
However, the required overlap window, flash timing and available acquisition modes are sensor-specific. A generic LED flash cannot be assumed to turn every rolling-shutter camera into a global-shutter camera.
Confirm hardware-trigger availability, trigger latency and jitter, exposure mode, strobe output, illumination timing, supported FPS and the interface used by the final host. Camera shutter type alone does not prove synchronization capability.
Technical reference:Teledyne Vision Solutions - Triggering.
The best shutter depends on what the application must measure or recognize, how the object moves and how the camera is mounted.
| Application | Recommended Direction | Most Important Validation |
|---|---|---|
| Stationary Document Capture | Rolling shutter is often sufficient | Resolution, focus, exposure and lighting |
| Static OCR Inspection | Rolling shutter may be practical | Smallest character size, sharpness and illumination |
| Moving Conveyor Inspection | Global shutter generally preferred | Object speed, allowable geometric error and exposure |
| Robotic Pick and Place | Global shutter for motion-sensitive positioning | Position repeatability, synchronization and motion blur |
| Barcode on Moving Belt | Evaluate global shutter with controlled light | Code element size, exposure and motion distortion |
| Mobile Robot Navigation | Depends on motion and algorithm | Camera movement, visual geometry and time alignment |
| Precision Measurement During Motion | Global shutter generally preferred | Dimensional tolerance, calibration and exposure timing |
| General Smart Terminal | Rolling shutter often sufficient | Image quality, lighting and cost |
These are engineering starting points rather than mandatory shutter selections. A fast rolling-shutter sensor, controlled motion and carefully synchronized illumination may be suitable in some dynamic applications. Final approval requires testing against the intended motion and measurement conditions.
A common selection mistake is to treat global shutter as a universal upgrade. In reality, the better sensor depends on the complete imaging requirement.
Rolling-shutter designs may offer attractive options for high resolution, low-light performance, cost and power consumption. But these advantages depend on the specific sensor generation and pixel architecture, not only on the shutter name.
The inspected object is stationary during acquisition.
The object moves slowly enough that readout distortion stays within the accepted tolerance.
High image resolution or fine texture matters more than very precise motion geometry.
The system has controlled lighting and camera motion.
The available rolling-shutter sensor provides a better overall match for power, cost or enclosure size.
A tested correction or illumination strategy meets the application requirements.
Sensor architecture should be selected after comparing actual candidate datasheets, not by assuming every global-shutter sensor is noisier or every rolling-shutter sensor is cheaper.
Industrial and embedded camera systems often operate under LED lighting. If the illumination intensity changes rapidly during frame capture, the timing relationship between the light and sensor can create image artifacts.
With rolling shutter, different rows may sample different phases of a flickering or pulse-width-modulated light source. The result can be horizontal bright and dark bands or inconsistent exposure across the image.
A global shutter removes the row-to-row exposure-time offset, but it does not guarantee that flickering illumination will never affect frame brightness. Entire frames may still be brighter or darker depending on the light phase and exposure timing.
For machine vision systems, stable illumination can be as important as the chosen shutter type. A technically suitable camera may still deliver inconsistent inspection images if the light source is poorly controlled.
A request for a "2MP global shutter camera" does not provide enough information for reliable sensor and module selection.
The required exposure, motion accuracy, interface and synchronization capability should be defined before the camera module is selected.
Object speed, camera movement, smallest feature, allowable position error and required working distance.
Resolution, frame rate, pixel size, exposure range, shutter mode, bit depth and available sensor readout timing.
External trigger requirements, acceptable timing variation, strobe duration and available lighting control.
USB or MIPI CSI-2 architecture, processor, operating system, driver requirements and camera control method.
Lens field of view, focus distance, depth of field, aperture and optical quality at the intended working distance.
PCB or FPC dimensions, connector, cable length, mounting constraints and available power.
Yes. Global shutter describes the image sensor's exposure architecture, while USB and MIPI CSI-2 describe different camera-to-host integration paths.
A global-shutter sensor can be incorporated into a compatible USB or MIPI camera design. However, available output modes, trigger features, firmware and software requirements depend on the exact module.
CK Vision lists sensor-dependent rolling-shutter and global-shutter options within its embedded camera portfolio. If your application requires controlled motion capture, review the MIPI CSI-2 camera module range or discuss suitable USB camera module configurations.
The USB interface generation does not determine sensor shutter type. A USB 3.0 camera can use a rolling-shutter or global-shutter sensor depending on its actual design. Always confirm the image sensor model and published shutter specification before approving a camera for moving-object inspection.
Datasheet specifications should be followed by real-image validation. The most useful test is not necessarily the one producing the most attractive sample image, but the one that reproduces the actual production task.
Capture a stationary target to evaluate focus, resolution, distortion and image quality without additional motion variables.
Run the conveyor, robot or camera at the intended operating speed and compare object geometry.
Use repeatable lighting and selected exposure times to distinguish motion blur from readout-related deformation.
If the application requires precise capture timing, measure trigger response and repeatability on the actual camera configuration.
Validate frame delivery, processing latency and operating stability using the intended host processor and application.
When comparing two candidate sensors, keep the scene, lighting, object movement, lens field of view and evaluation method as consistent as practical. Otherwise, differences caused by optics or image processing may be incorrectly attributed to shutter type.
Global shutter captures all pixels during the same exposure interval, while rolling shutter exposes different image rows at different times. The time offset between rolling-shutter rows can distort moving objects or scenes captured by a moving camera.
No. A moving object can still blur during a global-shutter exposure if the exposure time is too long. Global shutter primarily avoids row-timing geometric distortion. Reducing motion blur may require shorter exposure and appropriate illumination.
No. Global shutter is generally attractive for moving-object measurement and motion-sensitive positioning. Rolling shutter can be suitable for stationary or slow-moving applications, particularly when resolution, sensor availability, cost or image quality is more important.
It depends on readout timing, object motion and permitted measurement error. A fast rolling-shutter sensor may be suitable when the expected deformation is below the application's tolerance. Controlled strobe lighting may also help on compatible systems.
Not necessarily. Global shutter and hardware triggering are separate capabilities. A camera may capture global-shutter frames without exposing a dedicated external trigger interface. Check the complete module specification when synchronized acquisition is required.
Not automatically. A faster sensor readout often reduces rolling-shutter artifacts, but output FPS alone does not specify the time difference between relevant rows. Review the sensor's row timing and the actual operating mode.
Yes. Global-shutter CMOS sensors can be integrated into compatible MIPI CSI-2 camera modules. The selected image sensor, data format, lane configuration, driver and host processor must support the required operating mode.
Provide the application, object speed, acceptable position error, working distance, field of view, resolution, frame rate, exposure requirement, lighting method, trigger requirements, host processor, interface and mechanical dimensions.
For moving-object inspection, robotic positioning and other applications requiring consistent image geometry, global shutter is often the preferred starting point. For stationary capture or applications where rolling-shutter distortion is acceptable, rolling shutter may provide a better overall balance of resolution, image quality, power, cost and available sensor choices.
The shutter is only one part of the imaging design. Exposure time determines how far an object can move while light is collected. Sensor readout timing affects rolling-shutter deformation. Lighting, trigger control, optics, frame rate and host processing determine whether the camera works reliably in the actual system.
Before production approval, test the proposed camera under real operating speed, lighting and host conditions. A correctly selected sensor and lens combination is more valuable than a shutter specification evaluated in isolation.
Basler - CMOS Rolling Shutter Cameras
Basler - CMOS Global Shutter Cameras
Basler Documentation - Electronic Shutter Types
Teledyne Vision Solutions - Triggering and Exposure Synchronization
CUSTOM EMBEDDED CAMERA DEVELOPMENT
CK Vision can evaluate camera-module configurations for embedded machine vision, robotics and industrial imaging projects. Shutter selection should be matched with the target motion, sensor performance, lens, lighting, interface and host system.
Send your application, object speed, working distance, field of view, required resolution and frame rate, trigger needs, processor model and mechanical constraints for an initial engineering evaluation.