Intelligent Recognition Method of Laser Image for Surface Temperature of Metallic Materials under Ambient Light

In this paper, a visible light image intelligent temperature measurement technology presented in our previously paper is used to research the laser assisted intelligent temperature measurement. Through the image of 532 nm laser reflection on the surface of a copper plate at room temperature, it was found that the gray value of pixels in the green channel is negatively correlated with temperature, verifying that the features of a single frequency visible light image can reflect temperature changes. Combining the laser image features with the ambient light image features under multi-wavelength, non-normal incidence and multi-angle visible light environment effectively improves the training effectiveness of the machine learning model, and the temperature recognition error is reduced by 75% at most. The above research results provide theoretical support for intelligent temperature measurement methods using visible light images based on thermoreflectance technique and provide a new solution for application scenarios with high temperature measurement accuracy requirements.