Comparison of two algorithms based on mathematical morphology for segmentation of touching strawberry fruits

More than two strawberry fruits always contact with each other in the natural condition. The touching mature strawberry fruits have to be separated and their centers of gravity have to be found in order to be picked automatically by robot. Based on the mathematical morphological algorithm, two methods to solve this complexity were proposed, namely, Clustering Fast Segmentation and Watershed Region Segmentation. First, BP neural network algorithm was used to separate the strawberry fruits from the background. Second, several initial processes were applied to prepare for the following operation, including graying, threshold and holes filling. In the end, the close fruits were separated by these two methods, and then the gravity centers of the fruits were calculated. According to the comparison and analysis of the results of the two methods, it is proved that both methods can successfully separate the touching fruits and both of them have their own advantages and disadvantages, and they can be applied in different areas.