Imaged-based Modeling Of Hickory Growth

In the development of trees, the growth three-dimensional structure influences biophysical processes of photosynthesis and transpiration. At the forest level, three-dimensional structure plays a key role in the exchange of matter and energy between plants and the atmosphere, and affects biomass and micro-climate environment. An approach for reconstructing three-dimensional models from images of trees had received much attentions of researcher in computer graphics, thus it also could reproduce and preserve the growth process, which provides technical support and theoretical guidance for management of forestry. As the research object, digital images of 44 hickories were obtained in demonstration garden of Qingliangfeng town of lin'an. The images were captured around the hickories by high-definition video camera in summer and winter during the 5, 8, 11, 15 and 20 years growing period. They were analyzed by automatic camera tracking system (ACTS) to determine the camera parameters and the trees'three-dimensional points in the captured image. Imaged processing technology and computer programming language were used to reconstruct the hickories'three-dimensional model of development and growth. The results were as follows:1. Reconstructed the three-dimensional point model of hickories. Removed the noise points from analysis the summer images by ACTS, leaving only the trees'points. According to the camera perspective changes in principle, removed the noise points in the three-dimensional though selection these noise points which were corresponding to the two-dimensional image, and the three-dimensional point model of hickories'crown was got from removing noise points of each image. As capturing constraints of environment, it was found that the error was large or could not calculate when the point model of hickories were used to be estimated the height and crown width.2.? Camera parameters were obtained by ACTS to reconstruct branch model. Key points were selected in the hickory images, two keys points could determine a branch, and then the branch model was built in the man-machine interface. The branch model was mainly used to calculate the data which is difficult to manual measurement. The height of hickories was calculated by the benchmark, and the difference between the estimated height and measured height was detected in paired T-test. The P values were 0.580, 0.726, 0.489, 0.225 and 0.205 which corresponding 5, 8, 11, 15 and 20 years of hickories. As P>0.05, there were no significant between calculation and measurement. So the length of each branch could be calculated under the reconstructed branch model.3. The OpenGL framework was constructed to express the three-dimensional scene. Based on mouse interaction, the points model and branch model could pan, zoom, rotation, in which achieved observation purpose with different angles of the three-dimensional.