High-throughput plant phenotyping (HTPP) relies on the precise measurement of various morphological and physiological traits for cultivar development, enabling the selection of genotypes resilient to biotic and abiotic stresses. This process necessitates the cultivation and monitoring of numerous germplasm lines within a single crop in controlled field settings. The success of HTPP fundamentally depends on the precise spatiotemporal registration of remote sensing imagery to track plant growth and development accurately over time. However, achieving the requisite alignment is often impeded by the constraints of consumer-grade global navigation satellite system (GNSS) receivers and dynamic scene changes in agricultural environments. To address these challenges, we present a novel, robust, and cost-effective feature extractor and descriptor that leverages local field geometry for accurate image matching. This approach demonstrates exceptional registration accuracy, achieving a Root Mean Square Error (RMSE) of 2.68 cm on a multi-plot wheat field, even in the face of complex scene alterations. The resilience of our method to both geometric and photometric transformations underscores its potential as a promising solution for enhancing precision agriculture applications. By effectively overcoming the limitations of traditional methods, our feature extractor and descriptor pave the way for more accurate tracking of plant growth, ultimately contributing to the development of more resilient crop varieties. This advancement is crucial for precision agriculture, enabling high-accuracy monitoring and analysis of plant traits, which can significantly improve crop yield and sustainability. Our innovative approach provides a robust tool for researchers facilitating better-informed decisions in crop management and breeding programs.