Improved fluid search optimization-based real-time weed mapping

Abstract

In the field of agriculture, variable-rate herbicide spraying (VRHS) technology has been used to solve the low efficiency of pesticides and crop chemical residues. The key of VRHS is the quick and precise identification of weeds from field images, which forms a weed map. Fluid search optimization (FSO) was able to simplify the threshold optimization process to create a weed map, which simulated the fluid flowing from high pressure to low pressure, but it is time consuming and often converges prematurely. So, an explosion mechanism and a two-phase optimization were introduced to improve the FSO-based segmentation algorithm. Experiments of segmentation weeds from a corn field at seedling growth stage showed that the IFSO algorithm obtained the best accuracy of 93.3% and the least running time of 0.019 s, compared with the standard PSO, GA, and FSO algorithms.

Keywords

• Variable-rate herbicide spraying
• Weed map
• Fluid search optimization algorithm
• Otsu