Plant Propagation with the Aid of Image Analysis Techniques

In this work, we propose an efficient plant propagation system with the aid of image analysis techniques. A lateral bud detection algorithm is designed and applied to determine the cutting line for plant propagation. We further develop an automatic control system to cut the plant based on the recommended cutting line, and the overall system aims at reducing labor cost and increasing the effectiveness of cutting the plants in the propagation process. Three steps including plant segmentation, feature extraction, and lateral bud positioning are designed to achieve fully automatic lateral bud detection. The plant segmentation step is based on K-means clustering, and we design a cluster merging mechanism and a smoothing process to obtain better segmentation results. Skeleton-based and curvature-based features are then extracted for the lateral bud positioning step. We investigate into two different approaches of lateral bud positioning, i.e. the heuristic thresholding approach and the SVM-based learning approach. The corresponding precision and recall rates show that the SVM-based approach performs better.