Abstract:
Robotisation of forestry harvesting in China has the potential to achieve great productivity benefits. The paper presents a method which is to recognize tree trunks based on BP neural network. Firstly, the color marks and the training sample data of neutral network have been extracted. Then, the neutral network model is set up. The collecting sample data is used to train the neutral network and get feasible network weighted value. At last, the outline of the trunk is identified by Hough Transforms. We evaluate the method can satisfy the need to recognize the trunk properly by the simulating image based on the BP neutral network. This method lays a strong foundation for further research on autonomous operation of forestry harvesting robots in steep terrain.
