Three-Dimensional Reconstruction of Soybean Canopy Based on Multivision Technology for Calculation of Phenotypic Traits

Precise reconstruction of the morphological structure soybean canopy and acquisition plant traits have great theoretical significance practical value for variety selection, scientific cultivation, fine management. Since it is difficult to obtain all-around information on living plants with traditional single or binocular machine vision, this paper proposes a three-dimensional (3D) method reconstructing calculation phenotypic based multivision. First, multivision system Kinect sensor was constructed point cloud data in three viewpoints, different fertility stages as research object. Second, conditional filtering K-nearest neighbor (KNN) algorithms were used preprocess raw 3D cloud. The clouds matched fused by random sample consensus (RANSAC) iterative closest (ICP) accomplish canopy. Finally, height, leafstalk angle crown width calculated experimental results showed that average deviations 2.84 cm, 4.0866° 0.0213 m, respectively. determination coefficients between values measured 0.984, 0.9195 0.9235. deviation RANSAC + ICP 0.0323, which 0.0214 lower thanthe algorithm. enable precise quantitative detection traits.