Abstract: Aiming at the shortcomings of bolt looseness diagnosis methods for existing transmission tower flange connection joints, a novel anti-loosening bolt looseness diagnosis method using the QR code and vision-based technique is proposed in this paper. In order to ensure that each bolt has a unique identifier and locate the single bolt from the picture, a bolt numbering rule suitable for transmission tower flange connection joints has been established. The effects of different formats, shooting heights, lighting intensities, and perspective angles on QR code recognition rate are studied. In order to eliminate the perspective distortion in bolt images, the corner points of QR code are used to establish the homography matrix and perform the perspective distortion correction on bolt images. And the bolt looseness angle can be identified based on the nut edge recognition results. A prototype flange node of the transmission tower was used for experimental verification. The results show that the proposed method can effectively identify the loosening angle of anti-loosening bolts on flange plate. The recognition errors of looseness angles are within ±0.63 degrees, and the standard deviations are between 1.24 and 1.75 degrees. The maximum recognition error is 15 degrees using uncorrected bolt images. The proposed method significantly improves the measurement reliability and promotes the engineering applicability.