Spherical 3D citrus fruit point cloud completion and phenotype detection method

• Main Authors: WANG SHAODONG, LI SHANJUN, XU SHENGYONG, LIAO QINGXI, YI TONGZHOU
• Format: Patent
• Language: Chinese; English
• Published: 19.03.2024
• Subjects: CALCULATING; COMPUTING; COUNTING; IMAGE DATA PROCESSING OR GENERATION, IN GENERAL; PHYSICS

The invention discloses a spherical 3D citrus fruit point cloud completion and phenotype detection method. The RGB image is segmented through the deep learning instance segmentation model, and the target segmentation precision is improved. Meanwhile, incomplete point clouds caused by shielding are complemented, so that the accuracy and effectiveness of a fruit point cloud phenotype detection result are improved. A circle center and a rotation axis are determined based on the point cloud at the maximum transverse diameter, rotation preliminary completion is performed based on the rotation axis, and further completion optimization is performed based on a deep learning point cloud completion network model, and an end-to-end fruit point cloud completion model is established. The advantages of the depth image and the RGB image are fully exerted, and a new thought and method are provided for the field of fruit point cloud completion and phenotype detection. The technical scheme has the advantages of high efficiency