Abstract:
Objective In order to improve the defect detection capability of large-sized specimens, line scanning thermography has become an effective non-destructive testing method for building, metal, aircraft components and carbon fiber reinforced composite materials. The current line scanning thermography method is difficult to provide intuitive detection results for specimens with complex morphologies. Therefore, the three-dimensional thermography integrating detection specimen surface contour geometric data is receiving increasing attention. In order to meet the demand for non-destructive testing of non-planar carbon fiber composite plastic specimens with complex morphology, this paper proposes a system that uses only one thermal imager to perform active three-dimensional thermography on moving specimens, called monocular dynamic 3D active thermography. The system integrates the 3D contour of the specimen with an active thermography inspection, and is capable of inspecting composite specimens with complex geometries and giving intuitive thermography defect inspection results.
Methods This proposed monocular dynamic 3D active thermography system consists of a line laser, a thermal camera, an actuator, and a calibration board and does not require an additional 3D profiler (Fig.1). Compared with traditional 3D active thermography detection methods that require additional 3D cameras to obtain surface contour information of the detection object; This system does not require an independent 3D sensor, and the thermal imager acts as both a three-dimensional sensor and a temperature sensor. To achieve this function, based on laser joint line scanning thermography, this paper proposes a mathematical model that unifies the line scanning model and the pinhole camera projection model, and uses the line laser as the thermal excitation for active thermography and the spatial encoding for 3D reconstruction. The algorithm realizes 3D reconstruction, spatio-temporal reconstruction for nondestructive inspection, and the registration of the reconstructed thermogram sequence and the 3D point cloud (Fig.2).
Results and Discussions Experimental system of monocular dynamic 3D active thermography used FLIR 6702and a 20 W laser with a center wavelength of 808 nm is established (Fig.3). The standard height specimen calibration experimental results show that, in the height measurement range of 1 mm to 150 mm, the average error is 0.16 mm, and the maximum error does not exceed 0.25 mm (Fig.4). The carbon fiber intake pipe experiment shows that the method has the ability to reconstruct the 3D temperature field of large-size carbon fiber composites and detect the defects (Fig.5). Compared with other 3D thermography techniques, the advantage of the proposed system is its simple structure, which is a single camera, single excitation 3D thermography system without the need for additional 3D measuring instruments or complex feature extraction and matching algorithms. The limitation lies in the fact that only laser can be used as both a thermal excitation for active thermal imaging and a spatial encoding for 3D reconstruction. It can only achieve active thermal imaging with laser excitation and has certain limitations on the wavelength distribution and power of the laser. Therefore, the proposed active 3D thermal imaging technology is specifically designed for the detection of carbon fiber composite.
Conclusions The monocular dynamic 3D active thermography system is proposed in this paper for 3D measurement, defect detection and 3D temperature filed measurement. Based on the joint line scanning thermography, this paper achieves three-dimensional measurement function by the jointly calibration of the inspected object, laser, and infrared camera. With only one calibration board added, this paper proves that thermal imaging can achieve three-dimensional measurement and the experiment distance measurement error is controlled within 0.25 mm. The proposed system works fast, robust, and can be used for quality control on the production line.