Abstract: By using computer image processing technology, features of weed in the corn seedling field were analyzed, the field weed was recognized from the corn seedling, the weed position and growing state were found out. In the research, the two-apex method was used to filter soil background, the weed was recognized depending on projective area, leaf length and width, weed growing state was confirmed according to weed projective area, weed position was calculated depending on projective area center. The experiment showed that the method could identify acerose endogen weed correctly.