CONNECTED TECHNOLOGY. CLEARLY EXPLAINED.
Machine vision cameras
Choose machine-vision cameras as part of an inspection system: object detail, optics, lighting, motion, image transfer, processing and acceptance evidence.
A machine-vision camera captures an image for a defined inspection or measurement task. The result depends on the object, optics, lighting, timing and processing around it. Begin with the smallest relevant detail and the decision the system must support, rather than treating a camera resolution or an AI feature as a complete inspection system.
Define the inspection before the camera
Describe the decision
Specify what the system must distinguish, count or measure, and how an authorised reviewer establishes the reference result. Include acceptable and unacceptable examples, natural variation and uncertain cases. A collection of visually appealing demonstration images does not define an inspection acceptance criterion.
Describe the scene and motion
Record the object size, viewing area, working distance, movement and available mounting space. Include material, finish, orientation and environmental changes. A stationary sample on a bench and a moving item on a production line present different acquisition conditions.
Describe the full system
List camera, lens, illumination, mounting, trigger source, transfer interface, processing hardware and software. Identify the responsible integrator and operating owner. Check the exact supported combination and the required maintenance rather than evaluating each headline specification in isolation.
Build the image-to-decision brief
Lens, lighting and useful detail
Basler’s lighting guidance explains that illumination choices reveal different object features and can change the suitability of an inspection image. Ask the integrator to demonstrate the intended lens and lighting with representative surfaces and positions. More pixels alone cannot recover detail that the optical and lighting arrangement does not capture usefully.
Shutter behaviour and movement
Basler distinguishes global exposure, where sensor pixels begin and end exposure together, from rolling exposure with a row-to-row time offset. Its documentation notes possible rolling-shutter distortion during fast movement. Check the actual sensor mode and application; a global-shutter label does not remove every source of blur, distortion or timing error.
Exposure and trigger timing
Define how the relevant item is associated with its image and result. Review supported trigger sources, exposure behaviour and any illumination synchronisation with the integrator. Keep acquisition timing separate from the time needed to transfer, process and act on the image. Model-specific trigger support needs checking in the camera documentation.
Interfaces, throughput and records
Compare the complete acquisition-to-processing path, including supported host hardware, drivers, software versions and storage. Ask how dropped, duplicate or incomplete images are identified. A link-speed or advertised frame-rate figure does not establish sustained throughput with the intended resolution, exposure, processing and retention settings.
Algorithms and representative evaluation
Keep image acquisition and the inspection algorithm as distinct parts of the review. Use representative acceptable, unacceptable and borderline samples with a documented reference assessment. Report missed defects, false rejects and unassessed conditions explicitly; a model confidence value is not itself a measured defect-detection rate.
Maintenance and controlled changes
Plan cleaning, focus checks, lighting replacement and the handling of changes to product appearance or line conditions. Retain acquisition settings, software versions and reviewed sample results. Treat a changed lens, exposure, model or decision threshold as a change to evaluate, rather than assuming the original acceptance result still applies.
An illustrative acceptance review
Ask the authorised integrator to trace a representative item through its image, timestamp and inspection result. Review accepted, rejected and uncertain examples, along with a missing-image condition. This is a proposed review; IoTPapa has not inspected a production line, calibrated a vision system or measured defect-detection performance.
Primary acquisition and optics references
Basler: lighting in machine vision, Basler: electronic shutter concepts and Basler: model-specific triggered acquisition provide manufacturer explanations. They do not establish that a particular camera-and-software combination meets a site’s inspection requirements.
Connect images with the wider operation
Manufacturing connects the inspection decision with process ownership. On-device intelligence explains deployment and evidence for local processing. No exact-model machine-vision camera is presented here as independently tested.
By Admin · Reviewed 18 October 2026. Documentation-based product-type guide; no line trial, calibrated dimensional measurement, image benchmark, verified defect-detection rate or safety function claimed. Return to Industrial IoT.
Connect inspection, processing and maintenance
EXPLORE FURTHER
Manufacturing
Place inspection within a responsible operating process.
EXPLORE FURTHER
On-device intelligence
Review local processing, model limits and evidence.
EXPLORE FURTHER
Condition monitoring and maintenance
Connect observations with investigation and action.