Vision and Endoscopy
Unlock advanced vision and endoscopic image processing
Our vision modules are designed for medical and industrial applications, accelerating development with high-performance, product-grade features and intuitive visual prototyping tools—enabling seamless integration into your workflow.
- 3D vision processing
- Endoscopic video analysis
- SLAM and computational geometry
Endoscopic 3D Scene Understanding
Transform endoscopic video into 3D reconstructions for measurement, pre-op segmentation registration, and local mapping. Convert monocular video to 3D.
Shape Matching & 6D Pose Estimation
Register surfaces and volumes, determine the pose of objects from images for endoscopic or industrial applications.
Computational Geometry
Process and optimize meshes and point clouds with advanced filtering and transformation tools.
Demos
Endoscopic Stereo SLAM
This video shows our endoscopic SLAM pipeline on stereoscopic video of a laparoscopic screening procedure.
The stereo input is converted to a depth map using a state-of-the art learning based stereo depth estimation network. The resulting color + depth stream is then fed into our surfel-based SLAM pipeline.
Surface Registration
This video shows how to register a 3D surface scan to a CT or MR scan for interventional navigation using ImFusion Computer Vision Suite.
Endoscopic Tissue Tracking
This video shows how different features, including stereo depth reconstruction, optical flow computation, feature tracking and Holoscan integration come together to enable innovative medical applications. For this video we specifically worked with our clinical partners from Orsi to track a pre-op kidney organ segmentation in the endoscopic video as the camera is moved and the tissue interacted with.
Features
Vision
- Camera Calibration (Single, Stereo, RGB-D, Handeye, automated image selection)
- Marker Detection
- Stereo Reconstruction
- Native Stereo data and stream support including easy working with interlaced and side-by-side video
- Feature Detection and Matching
- Surface Registration: ICP, MAICP, Pose Graph Optimization
Endoscopy
- Tool segmentation
- Tissue Tracking
RGB-D/SLAM
- RGB-D SLAM algorithms
- Stereo SLAM algorithms
- Monocular SLAM algorithms
- RGBD Sensor support (Orbbec, Kinect, RealSense)
- Recording and playback of RGB-D streams
Computational Geometry
- Surface Reconstruction from point cloud data
- Geodesic distance
- Mesh processing (cropping, filtering, smoothing, reduction, ...)
- Mesh collision detection
- Point cloud processing (filtering, ...)
VR
- Rendering to HMD and other OpenXR devices
- Stereo rendering
Looking for inspiration or support?
Our forum is already packed with some common answers, tips and tricks and support information from our own product experts. Have a look or submit a request to our team.