How to get transform data with 3d point cloud？

[WeiWenQiang123456]

The point cloud is obtained by modeling using oblique photography.Due to the large number of point clouds, the point cloud is extracted by voxel segmentation through octree. Each voxel is the smallest research unit, which contains a lot of point clouds.

The voxel is equivalent to a picture and contains color, three-dimensional coordinates, roughness, and other spatial features. Each voxel is written in the database, and a row contains 89 features.In order to convert a row of one-dimensional data into a two-dimensional picture, since the maximum value is a color value of 256, it is transformed into a matrix of 256 by 256 by normalization and standardization, representing a voxel.

In IIC, we know that transforming data is important to us. My question is whether I should first extract features from the point cloud by color and rotation transformation or first extract features and then transform the matrix of 256 times 256.In IIC, we know that transforming data is important to us. My question is whether I should first extract features from the point cloud by color and rotation transformation or first extract features and then transform the matrix of 256 times 256.