1999 Volume 61 Issue 2 Pages 61-69
Weeding is an essential operation for maintaining the beauty of lawn fields such as golf course and garden. Since intensive chemical spray is not desirable, it is necessary that the weed area is discriminated from lawn area. However, both weed and lawn usually have similar green color in summer. A method using textural features extracted from an image was investigated for detecting weed area in this paper.
Three textural features, Contrast Angular Second Moment, and Inverse Difference Moment were extracted from 9 or 16 regions in an image with and without image smoothing. The results showed that the features extracted from weeds' size well-fitted segmented image area with image smoothing could discriminate weed regions from lawn regions in lawn field.