Canopy extraction using thermal images significantly affects temperature mapping and crop water status estimation. This study aimed to compare several canopy extraction methodologies by utilizing a large database of UAV thermal images from a precision irrigation trial in a peach orchard. Canopy extraction using thermal images can be attained by purely statistical analysis (S), a combination of statistical and spatial analyses (SS), or by synchronizing thermal and RGB images, following RGB statistical analysis (RGBS). A quantitative comparison of these approaches in the context of water status estimation for precision irrigation is currently absent from the literature. Our main objective was to compare six different canopy extraction methods representing examples of the three approaches defined above: 1) Quartile (Q) (S approach); 2) 2-pixel erosion (2PE) (SS approach); 3) 5-pixel erosion (5PE) (SS approach); 4) Edge detection (ED) (SS approach); 5) Thermal image masking with excess green (ExG) 70 threshold (EXG70) (RGBS approach); 6) ExG Otsu threshold (EXGO) (RGBS approach). Ten high-resolution UAV thermal images were acquired in a 4 ha commercial peach orchard in northern Israel during growth stage III of 2019. A confusion matrix was built per image for each canopy extraction method and the accuracy and kappa values were evaluated.  Additionally, an experiment was performed in part of the orchard where three plots were irrigated by 150, 100, and 0% (for a minimum of three weeks) of reference evapotranspiration (ET₀). Five trees were designated in each plot. A histogram analysis of these trees was performed, including the number of extracted canopy pixels in each plot. The degree to which canopy temperature was affected by the different extraction methods and the subsequent effect on the crop water stress index of the designated trees was assessed. The canopy extraction methods yielded substantial differences. The kappa value ranged from 0.55 (EXG70) to 0.91 (EXGO). The number of canopy pixels in the stressed 0% ET₀ plot ranged from 1492 (5PE) to 5251 (EXGO). These differences can directly affect the accuracy of water status predictions. The precision agriculture toolbox could highly benefit from the above comparison in order to advance decision-making capabilities in complex orchard systems.