Towards low-cost contactless high-temperature estimation based on colorimetrie fusion

Accurate high-temperature measurement is very important for process monitor of the industrial system. Because the temperature of industrial hot object may be several thousand centigrade, for example, welding molten pool, the measurement range and the high price limit the application of traditional high-temperature measurement in the field. According to the colorimetric theory, we propose a low-cost contactless sensor fusion method for estimating the high-temperature of hot object. The proposed method adopts the ordinary camera and the filters to obtain the images of high temperature object at different wavelengths. Then, the nonlinear partial least squares is adopted to predict the temperature based on the gray values of the images. The proposed method maps the input space to the high dimensional space and uses the parameters of the prediction model are estimated by the iterative optimization. The proposed method could deal with the high correlation between inputs to ensure the generalization of prediction model. Since the temperature of the filament of a incandescent lamp can reach several thousand centigrade, the filament images at different voltage obtained in our test platform are used in the experiments. The results verify that the proposed method has higher effectiveness and can be applied for the high-temperature measurement correctly.