Segmentation of green vegetation of crop canopy images based on mean shift and Fisher linear discriminant

In this paper, a hybrid method of combining the mean shift (MS) with the Fisher linear discriminant (FLD) is implemented to improve the performance of crop image segmentation. The highlight is the adoption of a point-line-distance-based strategy for weighting training data at the stage of the FLD. A wide set of images was employed to test the proposed method, and the results demonstrate its high quality and stable performance. In addition, the simulation results show that the point-line-distance-based strategy takes affect via enlarging the distance of class means, increasing the between-class scatter while decreasing the within-class scatter.