[photo] Steel frame of a car on a car assembly plant

Case StudyAutomotive suppliers: Company A

Appearance inspections that require a lot of vision systems. Fluctuations in inspection accuracy. Learn how they solved the problems unique to working with large parts.


Company A provides parts to a major automobile manufacturer, with an established record for large components. They considered building a more accurate and efficient appearance inspection system in line with the recent demand for higher quality standards and shorter delivery periods. However, they had the problems unique to large parts.


  • Variations between systems is a concern as multiple vision systems will be simultaneously be controlled.
  • Purchasing many vision systems could be cost.

After the introduction of "4D High Resolution" lenses

  • Variations between systems were suppressed by using our lenses, from which variations have been eliminated.
  • The number of cameras was reduced by using lenses that have a wide field of view and uniform resolution performance.

Issue 1: To capture uniform images with more than 20 vision systems

Problem 1: Variations in captured images due to differences in lens performance and the man-hours required to correct such variations

Large components manufactured by Company A are to be attached to various parts of an automobile. The shapes and sizes vary depending on where they will be attached and the type of automobile. In order to carry out appearance inspections efficiently, they initially planned to inspect components by simultaneously shooting the entirety of each component. Including left, right, top and bottom or other angles in some cases, using multiple vision systems. For the inspection of large parts, the inspection system needed more than 20 vision systems. However, the lenses they used varied in performance, causing variations in the results of the image inspections. They took time repeatedly making adjustments to obtain consistent defect detection ability, but their efforts were to no avail.

[photo] Empty body frame of the inside of a vehicle

Issue 2: To reduce cost by decreasing the number of vision systems

Problem 2: A uniform image cannot be obtained when the shooting area is expanded.

Building a new inspection system that suppressed differences in performance among vision systems and at the same time reducing the management man-hours and cost by reducing the number of vision systems was the goal. However, in order to inspect large parts in the same way as before with fewer vision systems, it was necessary to expand the shooting area using a wide-angle lens. But that's not all. This also required high performance lenses capable of detecting the smallest irregularities in the edges of the image frame.

The keys to solving these issues were...
Elimination of variations in resolution performance between lenses and expansion of the shooting area.

Solution of Issue 1: Adoption of lenses with little variation between individual lenses

In inspections using multiple vision systems, the inspection results will vary unless all the systems demonstrate the same performance. In particular, lenses need special attention as they tend to show differences in quality. Company A really wanted quality-assured lenses that do not have any individual differences. All Fujifilm "4D High Resolution" lenses for machine vision cameras are strictly checked by inspection devices we developed ourselves in order to suppress lens-to-lens variations in terms of performance. In addition, all lenses are given serial numbers for thorough quality control of each lens. They are also capable of suppressing resolution degradation when the shooting distance or aperture is changed, and successfully reduced variations in the image inspection results in the case of Company A, where the cameras are installed at several locations with different conditions. As these lenses deliver uniform performance, the time it takes to adjust the inspection system can also be shortened, easing the burden on workers.

  • * This story was arranged based on actual facts.