agentsclimarketplace

Matlab point cloud file io

Skill matlab/matlab-agentic-toolkit/skills-catalog/image-processing-and-computer-vision/matlab-point-cloud-file-io

The MATLAB Agentic Toolkit brings trusted MATLAB capabilities to AI agents, making engineering and scientific workflows agent-ready.

Install
npx -y skills add matlab/matlab-agentic-toolkit --skill matlab-point-cloud-file-io

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.

What its author says it does

Copied from the file, not written here

Read and write 3-D point cloud data using Lidar Toolbox file I/O. Covers PLY, PCD, LAS/LAZ, PCAP (Velodyne/Ouster/Hesai), E57, and IDC (Ibeo) formats. Use when loading point clouds from disk, saving to disk, choosing the correct reader or writer for a file format, extracting or preserving lidar point attributes, reading Ibeo IDC sensor recordings, or converting between formats.

The file declares its own license as MathWorks BSD-3-Clause. That is the author’s claim about this one file, and it is not the same thing as the license GitHub reports for the repository, which is listed with the other numbers below.

SKILL.md

13.8 KB, ~3.6k tokens by cl100k_base, as published. Nobody here has run it

Point Cloud File I/O

Read and write 3-D point cloud data in MATLAB using Lidar Toolbox file I/O functions, covering PLY, PCD, LAS/LAZ, sensor PCAP, E57, and IDC (Ibeo) formats.

When to Use

  • Loading a point cloud file from disk into a pointCloud object
  • Saving a pointCloud object to disk in any supported format
  • Deciding which reader or writer function to use for a given file format
  • Reading LAS/LAZ files with selective filtering (ROI, classification, GPS time)
  • Extracting or preserving lidar point attributes (classification, GPS timestamps, scan angle)
  • Reading Velodyne, Ouster, or Hesai PCAP sensor recordings frame-by-frame
  • Reading multi-scan E57 files with indexed access
  • Reading Ibeo IDC sensor recordings with message-based access
  • Converting between point cloud formats (e.g., LAS to PLY, PCD to LAZ)

When NOT to Use

  • Streaming live sensor data in real time (use velodynelidar, ousterlidar, sicklidar)
  • Processing or filtering point clouds after reading (use pcdownsample, pcdenoise)
  • Reading or writing surface meshes (use readSurfaceMesh, writeSurfaceMesh)
  • Saving point cloud variables to MAT-files (use save)
  • Visualizing point clouds (use pcshow, pcplayer, pcviewer)

Must-Follow Rules

  1. Route by file extension, not by habitpcread ONLY supports .ply and .pcd files. For .las/.laz use lasFileReader + readPointCloud. For .pcap use the sensor-specific file reader (velodyneFileReader, ousterFileReader, or hesaiFileReader). For .e57 use e57FileReader. Calling pcread on a LAS or PCAP file produces an error, not a warning. Similarly, pcwrite ONLY writes PLY and PCD — for LAS/LAZ output use lasFileWriter + writePointCloud.

  2. PCAP file readers require a mandatory device identifier — never guess itvelodyneFileReader requires a DeviceModel string (e.g., "HDL32E"). hesaiFileReader requires a DeviceModel string (e.g., "PandarXT32"). ousterFileReader requires a CalibrationFile path (JSON). If the user has not specified the device model or calibration file, ASK — do not assume a default. There is no default value; omitting it causes an error.

  3. Use two-output syntax to get point attributes from LAS/LAZ[ptCloud, ptAttributes] = readPointCloud(reader) returns a lidarPointAttributes object containing Classification, GPSTimeStamp, LaserReturn, NumReturns, ScanAngle, and more. The single-output form discards these attributes permanently. Intensity is on the pointCloud object (.Intensity property), NOT on lidarPointAttributes.

  4. Preserve attributes with three-argument writePointCloudwritePointCloud(writer, ptCloud, ptAttributes) preserves Classification, GPSTimeStamp, and other lidar attributes in LAS/LAZ output. The two-argument form writePointCloud(writer, ptCloud) discards all attributes. If the user has attributes from lasFileReader/readPointCloud, always pass them through.

  5. PLY only supports ascii and binary encoding — Calling pcwrite(ptCloud, "file.ply", Encoding="compressed") throws an error. Only "ascii" and "binary" are valid for PLY. The "compressed" encoding is exclusive to PCD format. Defaults (R2026a+): PLY="binary", PCD="compressed".

  6. Always use string syntax — Use double-quoted strings ("text") for all literal text arguments (filenames, device models, encoding values, Name=Value values). Use Name=Value syntax for all name-value pairs. Never use character vectors ('text') or legacy 'Name','Value' pair syntax.

Preflight Procedure

  1. List MATLAB functions to call
  2. Check references/INDEX.md for each (function-level + task-level tables)
  3. Read required quick-ref files
  4. State at response top: Preflight: quick-ref/xxx.md, quick-ref/yyy.md (or Preflight: none required)

Key Functions

FunctionPurposeFormatKey Constraint
pcreadRead point cloud from filePLY, PCDSingle call, returns pointCloud
pcwriteWrite point cloud to filePLY, PCDEncoding varies by format
lasFileReaderCreate LAS/LAZ reader objectLAS, LAZTwo-step: create reader, then readPointCloud
lasFileWriterCreate LAS/LAZ writer objectLAS, LAZTwo-step: create writer, then writePointCloud
velodyneFileReaderCreate Velodyne PCAP readerPCAPMandatory DeviceModel (2nd arg)
ousterFileReaderCreate Ouster PCAP readerPCAPMandatory CalibrationFile (2nd arg)
hesaiFileReaderCreate Hesai PCAP readerPCAPMandatory DeviceModel (2nd arg)
e57FileReaderCreate E57 reader objectE57Use readPointCloud(reader, idx) for multi-scan
ibeoLidarReaderCreate Ibeo IDC reader objectIDCUse readMessages to read scan data

Decision Framework

READING — What is the file extension?
├── .ply / .pcd → pcread(filename)
├── .las / .laz → lasFileReader(filename) + readPointCloud(reader)
├── .pcap (Velodyne) → velodyneFileReader(filename, deviceModel)
├── .pcap (Ouster)  → ousterFileReader(filename, calibrationFile)
├── .pcap (Hesai)   → hesaiFileReader(filename, deviceModel)
├── .e57           → e57FileReader(filename) + readPointCloud(reader, idx)
└── .idc (Ibeo)    → ibeoLidarReader(filename) + readMessages(reader)

WRITING — What output format does the user need?
├── .ply → pcwrite(ptCloud, "file.ply")
│         Encoding: "ascii" or "binary" (default: binary)
├── .pcd → pcwrite(ptCloud, "file.pcd")
│         Encoding: "ascii", "binary", or "compressed" (default: compressed)
├── .las → lasFileWriter("file.las") + writePointCloud(writer, ptCloud)
└── .laz → lasFileWriter("file.laz") + writePointCloud(writer, ptCloud)
           Optional: pass ptAttributes as 3rd arg to preserve attributes

PCAP sensor identification: If the user says "Velodyne", "Ouster", or "Hesai" — use the matching reader. If they just say "PCAP file" without specifying the sensor, ASK which sensor produced it.

Ibeo IDC files: If the user has an .idc file, use ibeoLidarReader. Unlike other readers, it uses readMessages (not readPointCloud or readFrame) and returns message-based scan data.

Velodyne device models: VLP16, PuckLITE, PuckHiRes, VLP32C, HDL32E, HDL64E, VLS128, VelarrayH800

Hesai device models: Pandar128E3X, Pandar64, PandarQT, PandarXT32

Gotchas

pcread called on LAS/LAZ file

pcread only supports PLY and PCD formats. Attempting to use it on LAS, LAZ, PCAP, or E57 files throws an error.

% WRONG: pcread does not support LAS
ptCloud = pcread("survey.las");

% CORRECT: Use lasFileReader for LAS/LAZ
reader = lasFileReader("survey.las");
ptCloud = readPointCloud(reader);

Compressed encoding on PLY file

The "compressed" encoding is only valid for PCD format. PLY supports only "ascii" and "binary".

% WRONG: compressed not supported for PLY
pcwrite(ptCloud, "output.ply", Encoding="compressed");

% CORRECT: use binary for compact PLY
pcwrite(ptCloud, "output.ply", Encoding="binary");

% CORRECT: compressed is valid for PCD
pcwrite(ptCloud, "output.pcd", Encoding="compressed");

Using pcwrite for LAS/LAZ output

pcwrite does not support LAS or LAZ format. It only handles PLY and PCD.

% WRONG: pcwrite cannot write LAS
pcwrite(ptCloud, "output.las");

% CORRECT: use lasFileWriter for LAS/LAZ
writer = lasFileWriter("output.las");
writePointCloud(writer, ptCloud);

Losing attributes with two-argument writePointCloud

When writing LAS/LAZ, the two-argument form discards lidar point attributes. Always pass the attributes object if available.

% WRONG: attributes are discarded
reader = lasFileReader("input.laz");
[ptCloud, ptAttr] = readPointCloud(reader);
writer = lasFileWriter("output.laz");
writePointCloud(writer, ptCloud);  % ptAttr lost!

% CORRECT: preserve attributes with 3-argument form
reader = lasFileReader("input.laz");
[ptCloud, ptAttr] = readPointCloud(reader);
writer = lasFileWriter("output.laz");
writePointCloud(writer, ptCloud, ptAttr);

Accessing classification from the wrong object

Classification, GPSTimeStamp, LaserReturn, and other lidar-specific attributes are on lidarPointAttributes, not on pointCloud. The pointCloud object holds Intensity (and Location, Color, Normal).

% WRONG: pointCloud does not have Classification
reader = lasFileReader("data.laz");
ptCloud = readPointCloud(reader);
labels = ptCloud.Classification;  % Error

% CORRECT: Use two-output syntax
reader = lasFileReader("data.laz");
[ptCloud, ptAttributes] = readPointCloud(reader);
labels = ptAttributes.Classification;
intensity = ptCloud.Intensity;  % Intensity is on pointCloud

Reading all PCAP frames without a loop

PCAP readers are frame-based iterators. There is no single function to load all frames at once.

% WRONG: No readAll or similar function
reader = velodyneFileReader("lidarData_ConstructionRoad.pcap", "HDL32E");
allPoints = readPointCloud(reader);  % readPointCloud is not a method

% CORRECT: Loop with hasFrame/readFrame
reader = velodyneFileReader("lidarData_ConstructionRoad.pcap", "HDL32E");
while hasFrame(reader)
    ptCloud = readFrame(reader);
end

Patterns

Read a PLY or PCD file

ptCloud = pcread("scene.ply");
fprintf("Points: %d\n", ptCloud.Count);
fprintf("X range: [%.2f, %.2f]\n", ptCloud.XLimits);

Read LAS file with classification filtering

reader = lasFileReader("aerial.laz");
[ptCloud, ptAttributes] = readPointCloud(reader, Classification=[2 6]);
labels = ptAttributes.Classification;
fprintf("Ground + building points: %d\n", ptCloud.Count);

Read Velodyne PCAP frames

reader = velodyneFileReader("lidarData_ConstructionRoad.pcap", "HDL32E");
fprintf("Total frames: %d\n", reader.NumberOfFrames);
while hasFrame(reader)
    ptCloud = readFrame(reader);
end

Read E57 file with multiple scans

reader = e57FileReader("building.e57");
fprintf("Scans in file: %d\n", reader.NumPointClouds);
for idx = 1:reader.NumPointClouds
    ptCloud = readPointCloud(reader, idx);
    fprintf("Scan %d: %d points\n", idx, ptCloud.Count);
end

Write point cloud to PLY (binary)

ptCloud = pointCloud(rand(1000,3), Color=uint8(rand(1000,3)*255));
pcwrite(ptCloud, fullfile(tempdir, "output.ply"), Encoding="binary");

Write point cloud to PCD (compressed)

ptCloud = pcread(fullfile(toolboxdir("lidar"), "lidardata", "highwayScene.pcd"));
pcwrite(ptCloud, fullfile(tempdir, "scene_compressed.pcd"), Encoding="compressed");

Read LAS and write to LAZ preserving attributes

reader = lasFileReader("input.las");
[ptCloud, ptAttr] = readPointCloud(reader);
writer = lasFileWriter(fullfile(tempdir, "output.laz"));
writePointCloud(writer, ptCloud, ptAttr);

Convert PCD to PLY

ptCloud = pcread("scene.pcd");
pcwrite(ptCloud, fullfile(tempdir, "scene.ply"), Encoding="binary");

Convert LAS to PCD

reader = lasFileReader("survey.las");
ptCloud = readPointCloud(reader);
pcwrite(ptCloud, fullfile(tempdir, "survey.pcd"), Encoding="compressed");

Read Ibeo IDC file

reader = ibeoLidarReader("sensor_data.idc");
fprintf("Message types: %s\n", strjoin(reader.MessageTypes, ", "));
fprintf("Total messages: %d\n", reader.NumMessages);

% Read all messages — returns array of pointCloud objects
ptClouds = readMessages(reader);

% Two-output form: get message metadata (timestamps, labels, plane info)
[ptClouds, messageData] = readMessages(reader);

% Filter by message type: "Scan" or "PointCloudPlane"
ptClouds = readMessages(reader, Messages="Scan");

% Filter by time range
tStart = reader.FileInfo.TimeStamps{1}(1);
tEnd = tStart + seconds(10);
ptClouds = readMessages(reader, Time=[tStart tEnd]);

ibeoLidarReader key details:

  • Supports IDC files from Ibeo FUSION SYSTEM/ECU sensors
  • Two message types: "Scan" (data type 0x2205) and "PointCloudPlane" (data type 0x7510)
  • readMessages returns an array of pointCloud objects (one per message)
  • Second output messageData is a cell array of structs with MessageType, TimeStamp, and (for PointCloudPlane) Label, ReferencePoint, PlaneOrientation
  • FileInfo property is a table with columns: MessageType, DataType, Description, NumMessages, TimeStamps
  • All properties are read-only

Conventions

  • Use tempdir for output paths in examples and tests to avoid permission issues
  • Default to Encoding="binary" for PLY and omit encoding for PCD (default compressed is best)
  • Always use double-quoted strings ("text") and Name=Value syntax — never character vectors or 'Name','Value' pairs
  • Include file extension explicitly in filename — format is determined from extension
  • PLY flattens organized M-by-N-by-3 to unorganized M-by-3; use PCD to preserve organized structure
  • NaN/Inf values are skipped when writing to PLY
  • lasFileWriter only supports unorganized pointCloud objects

References

Load when...Reference
Reading or writing LAS/LAZ files, need filtering NV-pairs, attribute details, or writer propertiesreferences/quick-ref/las-io.md
Reading PCAP files from Velodyne, Ouster, or Hesai sensorsreferences/quick-ref/pcap-readers.md

Copyright 2026 The MathWorks, Inc.

Keep looking

Skills are one crate of 328,083. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.