Multiple Baggage Identification Algorithm Based on Point Cloud Density Clustering

Abstract

The identification of multiple pieces of baggage during self-check-in of air baggage is an important part of suitability review. The paper focuses on the detection and discrimination of multiple pieces of baggage based on 3D point cloud data. Based on the classification principle of intraclass compactness and inter-class separation degree of pattern recognition, the point cloud density is calculated in the point cloud space. If the number of other point clouds whose Euclidean distance is less than a certain point is greater than the specified minimum threshold, Consider the point as a candidate cluster center, and determine the number of the final point cloud clusters in combination with the distance between the calculated candidate cluster centers being greater than a specified threshold. By analyzing the sample data of air passenger baggage system and combining with the actual situation of baggage handling lane, the algorithm limits the search range of the number of clusters to the upper bound, then selects the cluster centers and applies the density based point cloud clustering to baggage Of the multiple pieces of separation, and finally using K-means clustering of baggage data clustering separation, so as to determine the actual number of pieces of aviation baggage. The experimental results show that the air baggage point cloud data obtained a more good discriminant effect.