Methods and Systems for Fast Resampling Method and Apparatus for Point Cloud Data

A method for processing input point cloud having points, wherein each point includes a set of attributes including two dimensional (2D) and three dimensional (3D) coordinates and other attributes, comprising:

• sensing a scene via sensors, the sensors are in communication with a computer readable memory to produce the input point cloud;
• using a processor in communication with the computer readable memory, the processor is configured for:
• accessing the input point cloud;
• constructing a graph representing the input point cloud, based on each point in the input point cloud representing a node in the graph, and identify and connecting two neighboring nodes in the graph to obtain a graph edge;
• determining a graph filtering function based on the constructed graph;
• filtering each point in the input point cloud by selecting a subset of attributes for the points and applying by the graph filtering function on the selected subset of attributes, to determine at least one value for each point in the input point cloud,
• producing a probability for each point, based on the at least one value of the point compared to a total of all values of the points in the input point cloud, and a predetermined number of points in an output point cloud;
• sampling the input point cloud using random evaluation of the probabilities of each point, to obtain a subset of points in the input point cloud, wherein the subset of points is the output point cloud;
• and
• storing the output point cloud in the computer readable memory or outputting the output point cloud via the output interface in communication with the processor, wherein the output point cloud is used to assist in subsequent processing and assists in management of the input cloud data.

A non-transitory computer readable storage medium embodied thereon a program executable by a computer for performing a method, the method for processing stored input point cloud having points, wherein each point includes a set of attributes including two dimensional (2D) and three dimensional (3D) coordinates and other attributes, the method comprising:

• sensing a scene via sensors, the sensors are in communication with the non-transitory computer readable storage medium to produce the input point cloud;
• constructing a graph representing the input point cloud, based on each point in the input point cloud representing a node in the graph, and identify and connecting two neighboring nodes in the graph to obtain a graph edge;
• determining a graph filtering function based on the constructed graph;
• filtering each point in the input point cloud by selecting a subset of attributes for the points and applying by the graph filtering function on the selected subset of attributes, to determine at least one value for each point in the input point cloud,
• producing a probability for each point, based on the at least one value of the point compared to a total of all values of the points in the input point cloud, and a predetermined number of points in an output point cloud;
• sampling the input point cloud using random evaluation of the probabilities of each point, to obtain a subset of points in the input point cloud, wherein the subset of points is the output point cloud;
• and
• storing the output point cloud in the non-transitory computer readable storage medium or outputting the output point cloud via an output interface in communication with the computer, wherein the output point cloud is used to assist in subsequent processing and assists in management of the input cloud data.