A Real-time FPGA-based Image Processing System for On-The-Spot Decisions in Agricultural Field

Abstract

<b><sc>Abstract.</sc></b> High image resolution and fast processing speed with low power consumption are the current urgent needs of Precision Agriculture (PA) to make rapid decisions from field inspection to help prevent damages in the field. However, Field-Programmable Gate Array (FPGA), known as a cost-effective, reconfigurable, and parallel processor, has the acceptance for real-time object detection applications. Therefore, this study is focused on the integration of FPGA-based image acquisition and processing techniques to detect the color object and deliver the processed output in real-time. A Real-time FPGA-based Image Processing System (RFIP), which was developed and tested in the laboratory environment throughout this study. All the real-time data and reference data sets for the G-ratio and B-ratio algorithms were fit into a basic regression model, and a correlation of R²=99.56% was found between the system and the reference system. RMSE for detected pixel area was 1.25% of the total pixel area considered and Lin‘s concordance correlation coefficient (CCC) was 0.9873 with the confidential interval 95% (0.9852; 0.9893), which explains a better fit of the system to the reference system. The developed system showed the potential to meet the current demands of PA with improved imaging technology for agile real-time field monitoring applications.