Real-time precision crop identification in high weed-density environments for robotic weed control using spectral fluorescence imaging in celery

• Proposed a novel crop identification techniques to distinguished from high weed density fields. • A novel crop signaling method using Rhodamine B enhances crop visibility. • Proposed a custom illumination system for fluorescence visualization. • It has ability to overcoming occlusions in high dense weed organic farming. • Proposed method outperformed existing systems in speed and detection accuracy. This paper presented an a pioneering method for real-time robotic weed control in vegetable fields characterized by high weed densities, particularly in challenging scenarios where the foliage of crop plants was intricately intertwines with weeds, causing significant occlusion. This issue was particularly pronounced in organic farming environments where conventional classification algorithms often fell short. Our proposed method, termed crop signalling , involved the pre-transplantation treatment of celery crop plants with Rhodamine B (Rh-B), a fluorescent compound with unique optical properties that generated distinct signals readable by machines to effectively discern crop plants from weeds. A custom illumination system was specifically designed to excite the fluorescence properties of the Rh-B dye, facilitating visualization. Concurrently, a dedicated machine vision algorithm was developed not only to differentiate crop plants from weeds but also to accurately determine the stem locations of crop plants as they enter the soil, facilitating targeted weed-knife control. Experimental results showcased exceptional performance, achieving a 100% accuracy rate in detecting and distinguishing crop plants from weeds in densely populated fields, with no instances of false positives. The algorithm demonstrated a high precision rate of 99.66% in identifying celery plant stem locations across 313 images. This research represented a significant advancement in spectral fluorescence imaging for precision weed and crop classification, offering promising prospects for the application of robotic weed control in celery cultivation.