Research

A Computer Vision Approach to Automated Monitoring of Aeroponic Nozzle Operation

Galichenko Sergey A., Pikalov Evgeniy V.

Abstract

Reliable aeroponic irrigation requires monitoring not only hydraulic parameters but also the spray formed inside the root chamber. This study experimentally evaluates a lightweight computer vision approach for monitoring aeroponic nozzle operation from conventional smartphone video. A laboratory installation with five nozzles was investigated under a reference five-nozzle condition and a controlled reduced condition in which the same two nozzles were closed. The evaluated descriptors included background difference, reference-relative binary coverage, Sobel, Canny, Laplacian of Gaussian responses, and Laplacian-variance sharpness. The main data set contained 26,319 frames from 44 irrigation cycles. Cycle-level analysis showed statistically significant separation between the three- and five-nozzle conditions for all evaluated descriptors, with the largest observed effect size for binary coverage. A separate benchmark on a Raspberry Pi 4 showed that each processing branch was individually compatible with a target rate of 30 frames per second. The results demonstrate the potential of image-based monitoring as a complementary diagnostic channel while also identifying limitations related to reference calibration, fixed optical conditions, and the non-randomized experimental sequence.

Keywords

aeroponics · computer vision · nozzle monitoring · spray analysis · background subtraction · image processing