Weed Detection Based on the Fusion of Multiple Image Processing Algorithms

Crop weeding is the most time consuming part of agricultural production, which requires a large amount of manpower and material resources. At present, weeding operations are basically done by manual removal and manual spraying of pesticides. On the basis of this situation, an automatic weed recognition algorithm based on multi-image processing algorithm has been proposed in this paper to reduce the production cost of agricultural products and the environmental pollution caused by pesticides. Further, the algorithm of implementation steps can be presented in the following steps. First, the soil background is segmented from the image by image preprocessing. Then, the crops and weeds are classified by area threshold, template matching and saturation threshold respectively. Finally, the weights of the three image processing methods are assigned, and the precise identification and location of crops and weeds are realized by voting. In this paper, an experimental study has been carried out on soybean weeding in the field. Meanwhile, the results of the study show that the average error rate of weed identification is 18.18% and the accuracy of weed identification is 81.82%. At the same time, compared with the single method of area threshold, template matching and saturation threshold, the accuracy of weed identification based on voting weight increased by 12.83% on average. Further, the algorithm in this study can provide technical support for intelligent agricultural applications such as weeding operation by intelligent mobile robot.