| 主权项 |
1. An image processing system comprising:
a data repository, including at least one storage device on a computer system, configured to store image data comprising points in a point cloud; and an image processor and program code, which when executed perform operations on the image data, including: placing the image data into a three-dimensional mesh, comprising: placing points into a number of cells in the three-dimensional mesh based on three-dimensional locations of the points; and generating implicit geometry data for the number of cells using the points placed into the number of cells, implicit geometry data being one of a representation of points relative to each of the number of cells in the three-dimensional mesh and values based on the relationship of points located within the number of cells, wherein the three-dimensional mesh comprises metadata including attributes comprising one or more of population, distance, and validity; identifying vectors of the image data in the three-dimensional mesh, wherein each vector of the vectors comprises a corresponding vertical array of single cells in the three dimensional mesh; identifying clustering data using the vectors, wherein the clustering data comprises a plurality of clusters; assigning a label to each of the plurality of clusters based on a probability that a given cluster represents a given object; thereafter defining classes by one or more of the plurality of clusters, the classes comprising identifications of the given object; extracting feature vectors from the classes and comparing additional vectors to the feature vectors to determine corresponding classes for the additional vectors; and generating an image using the feature vectors and the corresponding classes for the additional vectors. |
