A fruit three-dimensional digitization method based on point cloud

The invention discloses a fruit three-dimensional digitization method based on point cloud, which comprises the following steps: 1) acquiring fruit point cloud data by using a portable laser three-dimensional scanner; 2) adopting non-local classification, interactive threshold denoising and decimation; 3) Using point-octree, MLS fitting to simplify the fruit point model; 4) extracting fruit slicesof that point cloud model through PCA attribute calculation and recursive increment of geometrical similarity; 5) Registering fruit point cloud models by oriented-splat team registration and hierarchical global registration;; 6) performing texture synthesis of fruit point cloud based on texton sub-sampling weighted template. This method reconstructs the real fruits in the growing process, and proposes some methods and algorithms such as denoising, extraction, registration and direct texture synthesis, which make up for the shortcomings of rough effect, low coding efficiency and large storagespace of direct texture sy