Disclosed is a circuit board fault infrared image automatic detection method. The image information of the working state of a circuit board is obtained by collecting a visible light image and an infrared image simultaneously. First, a visible light image and an infrared image of a circuit board working normally are collected and stored in a database. During detection of the circuit board which malfunctions, circuit board images (including a visible light image and an infrared image) taken are compared with the images under a normal state in the database, and a computer processes the images to find an area and a specific position where a component or line with abnormal temperature is located and automatically make a mark on the visible light image. Thus, automatic detection of a possible faulty component of the circuit board is completed. The technology can be applied to fields such as circuit board maintenance and equipment fault monitoring, and can also be applied to power equipment fault repair, petrochemical pipeline coke plugging, thermal insulation layer fall-off check, railway locomotive axle condition monitoring, fire prevention, flood control, disease prevention and other temperature change related non-contact measurement fields.