Bionic vision sensor optical flow prediction method based on hybrid neural network

• Main Authors: LV HENGYI, FENG YANG, ZHANG YISA, ZHAO YUCHEN, SUN MING, HAN CHENGSHAN
• Format: Patent
• Language: Chinese; English
• Published: 11.10.2022
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; IMAGE DATA PROCESSING OR GENERATION, IN GENERAL; PHYSICS

The invention discloses a bionic vision sensor optical flow prediction method based on a hybrid neural network, relates to the field of image processing, and solves the problems that the existing optical flow prediction method has a smear or target loss phenomenon, so that the optical flow prediction precision is influenced, the calculation amount is increased, and the advantage of low data volume of an event camera is lost. The method comprises the following steps: firstly, acquiring a public data set on a network, and adaptively slicing a spatio-temporal data stream; then constructing a hybrid neural network for optical flow prediction of the bionic visual sensor, so that the network can perform optical flow prediction; secondly, designing a loss function of overall network training, and performing supervised training on the network to obtain a supervised optical flow estimation model with higher speed and higher precision; and finally, performing optical flow prediction on the spatio-temporal data flow of