Weed Community Feature Recognition Based on NSST Transformation

摘要

Precision spraying of herbicides is a trend in the development of modern precision agriculture, but traditional image processing techniques focus on the identification of weeds in single plants or small areas. Aiming at this problem, this paper proposes the concept of weed community, and uses the weed community in the Chinese cabbage field as the object of this paper. The feature identification of weed communities in the image is analyzed by non-subsampled shearlet transform. The numerical value of the Chinese cabbage and weed community can be distinguished by simulation experiments, that is, the mean value is 0.6, and the correct recognition rate of the hybrid image of Chinese cabbage and weed can reach 58%. Compared with other traditional methods, this method has certain advantages. It can extract and distinguish the weed characteristics in the Chinese cabbage field to a certain extent, and realize the extraction of farmland weed community characteristics, which lays a foundation for subsequent research.