Profile Measurement
Our client required software to measure the width of panels and the positions of profiles on their production line. We developed a solution using an Nvidia Jetson AI computer and an 8k camera. The system operates by starting the image capturing when the camera detects a panel on the line. As the panel moves, the entire surface is captured from a top view with a resolution of 8000×6000 pixels and saved. After capturing, geometric and lens distortions are removed using specialized algorithms. Subsequently, images are processed through various algorithms to maximize the accuracy of the extracted information. The solution provides feedback to the production management software, which in turn notifies the operator whether the product meets quality requirements.
Example of profile analysis:

Our application operates on an intelligent and automatic image analysis principle, accurately detecting panel edges at pixel-level precision, and measuring their width and straightness in real-time. Additionally, the software monitors shadows on the panels to quickly detect possible shifts or defects on the production line. All measurements are presented through comprehensive statistical reports, providing precise insights into dimensional variations, tolerances, and panel straightness, enabling swift and precise responses. Thanks to NVIDIA CUDA technology, the system processes high-resolution images very rapidly, ensuring smooth and uninterrupted production. The software includes a flexible REST API, making it easy to integrate with existing IT systems and enabling automatic, real-time quality control.
Example of width analysis:

Our application has successfully overcome several complex technical challenges. Leveraging NVIDIA CUDA technology, the system can rapidly process high-resolution images, ensuring real-time quality control even under intense workloads. Intelligent edge-detection and shadow analysis algorithms enable precise determination of panel edge positions, even under challenging lighting conditions, significantly reducing false alarms and improving measurement accuracy.
Operator interface:
