Diagnostic technique for classifying the quality of circuit boards using infrared thermal image

In this paper, an advanced method for diagnosing the quality of the electronic circuit from a thermal image of the circuit board based on image processing and pattern recognition is described. The implementation of this diagnostic scheme requires the successful combination of several technological issues into a complete system that has the capability of diagnosing the quality of the important components in the observed system. An infrared window is first used to view the components within enclosures to avoid heat radiation and conduction effects, and then the maximum temperature of each component within the pre-defined area on the thermal profile of the electronic circuit is processed to reduce the processing data number. For further feature extraction, a principal component transform method is employed to extract the principle features of the measured thermal images. Finally, an adaptive neural-fuzzy inference system (ANFIS) is utilized to recognize the thermal patterns corresponding to different quality categories. In the experimental implementation, we take the thermal images of a switch mode power supply for seven quality categories and verify our design notion to diagnose its quality from these measured thermal images