An Accurate Strain Gauge Positioning Approach Based on Geometry and Color Features.

Background: Strain gauges commonly used in structural testing are manually pasted, usually with a 1–4 mm deviation between the actual pasted position and the theoretical position. The deviation leads to measurement errors, which are more pronounced in high strain gradient regions with stiffness discontinuities such as openings, reinforcements, and notches. Objective: This study aims to obtain the actual pasted position of strain gauges and thus improving the testing measurement accuracy. Methods: A non-contact strain gauge positioning method is proposed. Firstly, considering that the strain gauges have a regular shape, the irregular borderlines are filtered out based on the geometric features of a borderline image. Secondly, considering that the color of the strain gauge is significantly different from the test piece, and the color features within the borderline are extracted by clustering and compared with the strain gauges to complete the recognition. Finally, based on the epipolar geometry, the matching relationship of the strain gauges in different images is obtained, and the positioning is completed based on binocular vision according to the strain gauge recognition results. Results: For a simple plate, the average positional error of strain gauges is reduced by 71%. For a small square tube, the average error is reduced by 44%. For a large cylinder, the average error is reduced by 32.4%. Conclusion: The proposed non-contact strain gauge positioning method can obtain high precision strain gauge positions, which effectively improves the test measurement accuracy. [ABSTRACT FROM AUTHOR]