During the safflower filament harvesting process, the harvesting device plays a pivotal role in determining overall efficiency. This study addresses the current challenges in harvesting devices, such as their bulky structure and the poor cutting efficiency of the blades, by developing a bio-inspired harvesting blade modeled after the pharyngeal teeth of the grass carp. By focusing on the structure of the pharyngeal teeth of the grass carp, digital image processing techniques were employed to extract its geometric features. Regression equations were then formulated to accurately fit both the contour and tooth profiles. Based on these fitted results, a bio-inspired blade design was proposed. To evaluate the blade's performance, simulations of the cutting process for safflower filaments were carried out using LS-DYNA software. These simulations compared traditional blades, bio-inspired contour blades, and bio-inspired tooth-shaped blades. The preliminary simulation results suggest that the bio-inspired blade exhibits significant advantages in cutting safflower filaments. Further testing was conducted using a dedicated safflower filament cutting performance test platform to compare and analyze the influence of different contour curves and blade edge tilt angles on harvesting efficiency and damage rates. The experimental results revealed that when the blade edge tilt angle ranged from 10 degrees to 30 degrees, the bio-inspired tooth-shaped blade outperformed the bio-inspired contour blade, which, in turn, demonstrated superior performance compared to the traditional contour blade. Specifically, the bio-inspired tooth-shaped blade achieved an average harvesting efficiency of 99.83%, the bio-inspired contour blade 97.68%, and the traditional contour blade 96.51%. To further assess the impact of the bio-inspired tooth-shaped blade's design parameters on cutting performance, both single-factor and multi-factor experiments were conducted. The results showed that, with a rotational speed of 61.39 rpm, an intake air velocity of 2.44 m/s, and a blade edge tilt angle of 15.32 degrees, the harvesting device reached optimal performance, achieving a harvesting efficiency of 99.97% and a damage rate of 1.53%. Subsequent repeated tests under these optimal conditions yielded an average harvesting efficiency of 98.95% and a damage rate of 1.54%. The relative deviation between these experimental results and the response surface optimization outcomes was less than 2%, thus fully meeting the practical requirements for safflower filament harvesting. The findings of this study not only provide valuable insights for the development of intelligent safflower filaments harvesting robots but also offer a theoretical foundation for the design of bio-inspired blades in agricultural mechanization fields.