A High Precision Quality Inspection System for Steel Bars Based on Machine Vision

Area of Science:

• Construction Engineering
• Machine Vision
• Quality Control

Background:

• Manual steel bar inspection is costly, slow, and imprecise.
• Ensuring steel bar quality is critical for building safety.

Purpose of the Study:

• To develop a high-precision, machine vision-based quality inspection system for steel bars.
• To address the limitations of existing manual detection methods.

Main Methods:

• Developed sub-pixel boundary location method (SPBLM) and fast stitch method (FSM).
• Integrated five sensors (CMOS, level, proximity, voltage, current) with smartphone image/video capture.
• Real-time data processing and detection via a smartphone application (APP).

Main Results:

• Achieved high accuracy: 0.04 mm absolute error and 0.002% relative error for diameter and spacing.
• Demonstrated zero error in counting steel bars.
• Enabled simultaneous detection of diameter, spacing, and quantity.

Conclusions:

• The developed system provides accurate and efficient steel bar quality inspection.
• The system's unique features offer significant advantages over existing methods.
• The device and method are suitable for widespread application in construction site quality control.