Bolt looseness detection algorithm based on 2D color space extraction and 3D depth measurement
Ran Yang, Ziling Guan, Lu Jiang, Yijing Zhao · 2022 IEEE 17th Conference on Industrial Electronics and Applications (ICIEA) · 2022
This project mainly uses 2D and 3D image processing technology, and develops fault detection algorithm accordingly. The research object we focused on is bolt. In reality, as for the images processing, the shooting environment is complex with different shooting angles, and the change of light in the environment will change the RGB value of the same point. Target detection based on model training is used in the processing. The bolt fault detection is based on the anti-loose line detection model based on two-dimensional image processing. In order to reduce Error detection rate, the color threshold dictionary is introduced to get the optimal threshold, and the bolt depth detection model based on three-dimensional point cloud processing is introduced to jointly complete the fault detection of target faults.