Image processing software is not only for automating inspection. It is a foundation for stabilizing quality on the shop floor, reducing variation in judgment and easing the workload of operators. At Katotech, we carefully organize inspection processes and judgment criteria, and propose image processing systems that work in real production.
Interest is growing in inspection, character recognition, classification and anomaly detection that combine conventional rule-based processing with AI and deep learning. But not everything should be replaced with AI: the method should be chosen with stability, explainability and running costs in mind. We value using the right technology for each part of the task.
Main Services
- Automation of visual inspection, dimensional measurement, positioning and missing-part detection
- Image processing algorithm development using HALCON
- OCR, code reading and integration with traceability data
- Evaluation of AI and deep learning for classification and anomaly detection
- Connection design for cameras, lighting, PLCs and host systems
- Ongoing support from PoC and evaluation to production deployment and improvement
Where It Works Best
It is especially effective where inspection criteria vary from person to person, visual inspection is a heavy burden, inspection speed needs to increase, or images of defects should be recorded. It can be applied to a wide range of processes, such as electronic components, molded plastic parts, print inspection, packaging checks and post-assembly checks.
On the shop floor, it is also essential that the system keeps running without stopping. We design not only for accuracy but also for changes in lighting, camera misalignment, ease of operation and ease of maintenance, so the system stays easy to use after installation.
From PoC to Production
We first check the workpiece, the purpose of the inspection, the required accuracy, the cycle time and the site conditions, and work out which method is realistic. Then, through prototype evaluation and a PoC, we define the imaging conditions, the judgment logic, the screens needed and how to connect with the equipment. Designing for real operation from the start, rather than chasing accuracy alone, is the key to a system that lasts.
Our work does not end with development: we also handle fine-tuning and reviewing inspection conditions after installation. Beyond showing results on screen, we design for saving logs and images, linking with host systems and analysis for quality improvement.