Author(s)

Jinzhi Wang, Xiuhua Li, Yonghua Zhou

Corresponding Author

ABSTRACT

In order to more accurately establish the three-dimensional point cloud model of sugarcane, so as to more effectively analyze the structure of sugarcane. KinectV2 depth sensor was used to obtain sugarcane point cloud data from different perspectives. In order to solve the problem that the traditional Iterative closest point algorithm (ICP) is demanding the attitude of space, an improved method is proposed. Firstly, the point cloud was preprocessed to remove redundant background and noise, and then the initial registration and accurate registration are carried out. The Normal Aligned adial feature algorithm (NARF) was used to search the point cloud. The fast point feature histogram (FPFH) was used to calculate the feature vectors corresponding to the key points. The sample consensus initial alignment algorithm (SAC-IA) was used to find the key points with similar FPFH characteristics in the point cloud. The correspondence between points was constructed and the rotation translation matrix was calculated, Initial registration was completed by rotating translation point cloud. Experimental results show that the error of the proposed algorithm is 2.64cm lower than that of the ICP algorithm, which can effectively improve the robustness and accuracy of the point cloud registration.

KEYWORDS

Sugarcane; KinectV2; point cloud; Point cloud registration