Main Article Content
Fruit tree pruning is a necessary means to improve yield and longevity. The current pruning means are mostly manual pruning, resulting in low pruning efficiency. The identification and extraction of target branch images is a key technology for automated pruning of fruit trees, which has seriously restricted the development of automated pruning of fruit trees. To this end, this paper proposes a branch image filling and extraction algorithm based on peach tree pruning technology. The algorithm is based on the analysis of peach tree branches, leaves and environment to obtain the intrinsic connection of branch images, so as to segment the branch part of the image with the leaf noise and carry out the obscured branch filling process to achieve the repair of the obscured part of the branches and obtain a more complete peach tree branch model. The algorithm was found to have a minimum correct recognition rate of 82.13%, a maximum of 89.79% and an average of 86.65% through experiments. The key factor in reducing the recognition rate is the total number of recognition hyperpoints, which are new pixel points generated during the image processing. These new points do not affect the shape and number of branches, and therefore do not affect the judgement of pruning and the related pruning process. Therefore, the correct recognition rate of the proposed algorithm can meet the practical needs and can be used as a theoretical basis for branch identification and automated pruning of peach trees.