Abstract: To enhance the efficiency and accuracy of instrument verification and reduce the impact of manual operation, reading, and recording on verification results of alpha and beta surface contamination monitors, an automatic verification device was developed based on automation technology and machine vision algorithms. The device comprises a double-layer source-changing turntable, image training software based on machine vision technology, and automatic verification control software written in C#. The verification process, hardware structure, and software were optimized, and an algorithm for identifying abnormal results was incorporated into the software. This algorithm can perform conditional filtering and abnormal data elimination based on the characteristics of the target recognition area, improving the accuracy of optical character recognition (OCR). Performance tests, including recognition rate testing, background influence testing, comparative testing, and automatic verification process testing, were conducted. Results show that the device achieves a 100% recognition rate for original data, with no additional background interference from the centralized placement of planar sources within the device. The maximum relative deviation of measurement results between manual positioning brackets and the automatic verification device is -6.0%, showing good consistency within the uncertainty range. While meeting the requirements of JJG 478-2016, this device optimizes radiation protection and inherent source safety, significantly improving verification efficiency.