Method of Multi-feature Fusion Based on SVM and D—S Evidence Theory in Weed Recognition

Abstract:

According to the low accuracy and low stability of the single feature-based method for weed recognition, a multi-feature fusion method based on SVM and D—S evidence theory was proposed. Firstly, three types of visual features such as color, shape and texture were extracted from the plant leaves after a series of image processing. Then, the plants were classified according to each type of features utilizing SVM and the results were used as evidences to construct the basic probability assignment (BPA). Finally, using D—S combination rule of evidence to achieve the decision fusion and giving final recognition results by classification thresholds. The experimental results show that the accuracy of multi-feature fusion method is over 97% and it has good performance on accuracy and stability compared to the single feature-based method in weed recognition.