Processing Point Clouds

Processing the complete point cloud including this excessive data is inefficient and leads to wastage of compute cycles. To alleviate this problem, we can use some simple techniques called "point cloud filtering" to remove the additional data points. Filtering not only removes useless and excessive data but also the data that is deemed adversarial for a given application. The following are some of the most commonly used filters from the Point Cloud Library. These filters are:

• VoxelGrid Downsampling Filter

• ExtractIndices Filter

• PassThrough Filter

• RANSAC Plane Fitting

• Outlier Removal Filter

While some of these are actual filters that use an underlying algorithm to improve the quality of your point cloud data, others are simple tools that are required at certain stages in your perception pipeline to aid in the extraction of a subset of the input point cloud.