An intelligent system control method based on visual sensor

In order to solve the complexity problem caused by the uncertainty of control system models, this paper utilizes visual sensors and intelligent control technology, and uses data-driven machine learning algorithms to extract representations from the original video for video representation learning, providing crucial semantic features for related tasks. Convolutional Neural Network (CNN) greatly improves the utilization efficiency of visual data and model performance, and realizes the recognition of complex application scenarios. In this paper, an intelligent system control method of time-domain vision sensor is proposed. The proposed method locate, track and measure the speed of moving objects based on CNN and image acquisition device, Asynchronous Temporal Vision Sensor (ATVS). The experimental results show that our proposed algorithm has improved its overall performance through video feature learning and clustering. It not only pays more attention to video spatial information to enhance the discrimination ability of learned video representations, such as scenes and objects, but also improves the tracking performance of visual sensors under various interference attributes