Research On Applying Computer Vision Technique To Weed-Sensing

The chemical herbicide is the most effective way of weeding. However, much chemical herbicide are wasted and may cause pollution of water, air and soil because of its use for large area. One promising way of solving the problem is to study and apply the intelligentized variable herbicide machine. The intelligentized variable herbicide machine can sense the density and location of weeds, and spray herbicide according these. So it can alleviate the harm of weeds, reduce cost and protect environment effectively. Because it is a new research filed to apply computer vision for weed-sensing, there are not mature theories to support the weed-sensing technology. Although the intelligentized variable herbicide machine sprays herbicide according to the results of weed-sensing, the herbicide can not be compounded while spraying. The concentration of herbicide must be made before spraying according the condition of weeds. The leaves area of weed is the standard of making the concentration. In this paper a set of software is made including four modules such as image-processing mould based on computer vision. An equipment for measuring area of live leaves is developed. They can be used to measure and analyses all green leaves. A short time algorithm for outdoor weed-sensing is realized in this paper. It can provide controlling-signal for variable control, and a system of weed-sensing and variable control for the intelligentized variable herbicide machine is designed.