Abstract

Technologies such as identification, detection, and measurement based on multiple industrial cameras have developed vigorously. Most of these technologies require synchronized video signals from multiple cameras, thus a general method of synchronization of multiple industrial cameras has always been an important research direction. To reduce synchronization error and hardware cost of the existing multi-vision sensor synchronization methods, based on multiprocessing or multithreading techniques, this paper designs three synchronous capture schemes for different application scenarios. The proposed schemes are tested for an example system of three industrial cameras with the highest frame rate for 600FPS. For long-time synchronous capture with a low frame rate, a read–write structure is proposed based on multiprocessing, which makes the synchronization error lower than the reciprocal frame rate when the frame rate is lower than 200FPS. For short-time capture with a high frame rate, a read-storage-write structure is used based on multiprocessing, which makes synchronization errors be about 1 ms when the frame rate is lower than 600FPS (theoretically ~ 1000FPS). To relieve the memory usage pressure of the read-storage-write structure, a read-cache-write structure is proposed based on multiprocessing/multithreading, which makes synchronization error be less than 1 ms when the frame rate is lower than 600FPS (theoretically ~ 1000FPS). Experimental results show that the proposed method provides a promising accurate and low-cost solution for general industrial camera synchronization applications.

Full Text
Paper version not known

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call