Dynamic Detection of Infants’ Video Based on All-Programmable SoC in NICU

The traditional Neonatal Intensive Care Unit (NICU) is equipped with camera monitoring. Medical personnel can obtain the monitoring videos of newborn babies in real time, so that the corresponding care and treatment can be performed according to the actual situation. However, the medical personnel's energy is limited after all, and it is impossible to keep staring at the baby's surveillance screen all the time. Once the condition of the newborn occurs and the medical staff fails to detect it in time, it may lead to the occurrence of nursing accidents in the neonatal intensive care unit. When babies feel pain, in addition to crying, they will express their discomfort by shaking their arms, kicking their legs violently, arching their backs, etc. This paper proposes a method that integrates newborn video picture acquisition, infant over-amplitude motion detection, and an audio alarm system. The single Gaussian background modeling algorithm is used to detect the baby's large-scale movement. When the baby in the picture detects a large movement, the audio alarm system responds in time to notify the medical staff. The hardware platform on which this design relies is the Zynq System-on-Chip (SoC) development board, and the software platform uses the Vivado series design tools. This method can reduce the occurrence of nursing accidents in the neonatal intensive care unit, while freeing medical personnel from paying attention to the newborn's video at all times. Therefore, it greatly reduces the workload of medical personnel and improves work efficiency.