Abstract: Computer interlocking automatic testing can save the labor cost in the phase of software testing, while providing reliable interface layout data is the premise of realizing automatic testing. According to the upper computer interface display specification, this paper proposed a method for extracting information from the upper computer interface of computer interlocking software based on computer vision image analysis and text recognition technology. This method implemented station information extraction based on upper computer interface screenshots by mining quantifiable features of signal devices and function button entities, as well as the character features of their names in the interface. Taking the computer interlocking software of Lianjiangkou Station of China Railway Guangzhou Group Co. Ltd. as an example, the feasibility of this method is verified, which provides a new idea for automatic extraction of station yard information.