A Novel Method for Compensating Temperature Measurement Error Caused by Dust Using Infrared Thermal Imager

Abstract

The dust generated in industrial processes has a serious impact on the accuracy of infrared thermometry, which is also the main reason for the limited use of infrared temperature measurement methods in industrial applications. To reduce the influence of dust on infrared temperature measurement, this paper proposes a new method for compensating measurement error caused by dust. First, the source of temperature measurement error caused by dust is analyzed, and a compensation method, in which the dust transmittance is important but difficult to determine, is proposed based on the principle of infrared temperature measurement. Then, to solve the difficulty of determining dust transmittance, we define a spatial temperature-level co-occurrence matrix and a neighboring temperature-level dependence matrix that can be utilized to extract the infrared thermal image’s texture features affected by dust. Finally, by integrating stacked denoising auto-encoder with optimized parameters and support vector regression, a dust transmittance model is established to determine dust transmittance based on the extracted features. Experimental results indicate that the proposed compensation method can reduce the influence of dust on infrared thermal imager’s accuracy effectively.