Abstract: In order to reduce the herbicide on crops and stress on the environment, this paper proposes to use machine vision to identify weed networks, and destroyes network of weeds from betweenrow crop by scalefree networks method. The paper identifies the crop, soil and weeds by machine vision, builts and mappes scalefree networks of weed according to the role of allelochemics, researches weed networks of preferential attachment, growth and clustering, founds that error tolerance and attack vulnerability are generic properties of weed networks, and proposes a method to attack and destroy weed networks. Compared with existing methods, new methods accord with the principle of ecoeconomic management.

Key words: weed, betweenrow-crop, scale-free networks, machine vision, herbicide