image_common

The image_transport library provides a plugin-based architecture for publishing and subscribing to images. It allows applications to switch between raw and compressed formats transparently without changing the application code.

Key features include:

• Publishers: Publisher and CameraPublisher (for images with optional camera info).
• Subscribers: Subscriber and CameraSubscriber (supports transport hints).
• Filtering: SubscriberFilter for integration with message_filters.
• Utilities: republish node for converting between transports and list_transports to see available plugins.
• Support: Lifecycle node support and QoS overrides.

The camera_info_manager_py package provides a Python interface for managing camera calibration data in ROS 2. It is designed for Python-based camera drivers that need to load sensor_msgs/msg/CameraInfo from disk, save new calibration data received via the set_camera_info service, and serve that calibration to the ROS 2 system.

Key responsibilities include:

• Loading calibration files from filesystem or package paths.
• Saving updated calibration data.
• Providing a service to allow external tools to update the driver's calibration.

The camera_info_manager package provides a C++ class designed for camera drivers to manage sensor_msgs/msg/CameraInfo.

It automates three key tasks:

1. Lazy Loading: Automatically loads calibration data from a file or a package:// URL the first time it is accessed.
2. Saving: Saves new calibration data to the specified file when a sensor_msgs/srv/SetCameraInfo service call is received.
3. Service Advertising: Automatically advertises the ~/set_camera_info service, allowing calibration tools (like camera_calibration) to update the driver's calibration data at runtime.

Typical Workflow:

• The driver node instantiates a CameraInfoManager with a calibration URL.
• The manager waits for the first call to getCameraInfo() or isCalibrated() before performing I/O.
• The driver then publishes both the Image and the CameraInfo on every frame.

The library supports two human-readable formats:

1. YAML (.yaml or .yml): The standard format produced by the ROS 2 camera_calibration package and used by camera_info_manager. It contains fields such as image_width, image_height, camera_matrix, distortion_model, and projection_matrix.

2. Videre INI (.ini): A legacy format used by older Videre Design stereo cameras. The parser supports the subset of this format commonly used in open-source camera drivers.

If you need to synchronize image streams with other sensor data (like PointCloud2) using message_filters synchronization policies (e.g., TimeSynchronizer or ApproximateTime), use image_transport::SubscriberFilter.

image_transport::SubscriberFilter acts as a drop-in replacement for message_filters::Subscriber. Its primary advantage is that it automatically decodes compressed image streams before passing them into the message_filters synchronization chain, allowing you to work with standard sensor_msgs::msg::Image types even if the source is compressed.

The camera_calibration_parsers package provides C++ functions to read and write camera calibration data. For most use cases, you should use the high-level API provided in parse.hpp. This layer automatically detects whether a file is in YAML format (standard for ROS 2 camera_calibration) or Videre INI format (legacy) based on the file extension.

The API supports three types of operations:

• write: Serializes a sensor_msgs/msg/CameraInfo to a stream or file.
• read: Deserializes from a stream or file into a CameraInfo.
• parse: Deserializes from an in-memory string buffer.

// Use parse.hpp for automatic format detection
#include <camera_calibration_parsers/parse.hpp>

// Example logic (conceptual):
// camera_calibration_parsers::read(file_path, camera_info_msg);
// camera_calibration_parsers::write(file_path, camera_info_msg);

When you instantiate a CameraInfoManager, it automatically advertises a set_camera_info service relative to the node's namespace (or an optional custom namespace prefix).

This service allows external calibration tools to send new calibration data to your running driver. When the service is called, the manager:

1. Updates the in-memory CameraInfo.
2. Saves the new data to the configured calibration URL.

An image_transport plugin is a pair of pluginlib plugins—one publisher and one subscriber—that work together to negotiate how image data is serialized on the wire.

When a publisher uses a specific transport, it encodes the standard sensor_msgs/msg/Image into a transport-specific message type. The corresponding subscriber plugin listens to that specific message type and decodes it back into a standard sensor_msgs/msg/Image before passing it to the user's callback.

When using image_transport::CameraPublisher, the CameraInfo topic is automatically derived from the base image topic name by appending /../camera_info. This creates a sibling topic.

For example, if your base topic is /camera/image_raw:

• Image topic: /camera/image_raw
• Camera info topic: /camera/camera_info

You can programmatically compute this sibling name using the image_transport::getCameraInfoTopic function.