360° Surround View Development Tools
OEMs and Tier 1 suppliers engaged in the development of automotive surround view systems face numerous challenges. At Cogent, we have supported enough vehicle demonstrations to know how time-consuming the initial tasks of camera placement, calibration, and UI configuration are. That’s why we have developed our own suite of tools to accelerate the deployment of these systems.
Camera Location Tool
One of the most constrained design decisions in developing a surround view system is camera placement: Cameras must see the surroundings of the vehicle clearly, yet ideally they shouldn’t be seen to not disrupt the aesthetics of the vehicle.
Our camera placement tool can help! It provides an accurate geometric simulation of camera distortion and field of view (FOV). Cameras can be electronically placed in a CAD model of the vehicle to estimate occlusions and blind zones caused by the different parts of the vehicle.
- Accurate geometric simulation of camera distortion and FOV
- Estimation of occlusions and blind zones caused by vehicle
- Matching of display pixels to camera pixels to minimize aliasing
and interpolation artifacts
- Simulation of multiple vehicle load configurations
- Rear view simulation according to FMVSS 111 (check compliance with FMVSS 111)
- Reference coordinate system support for ISO 8855, VW 01055, J670
- Import of 3D vehicle model in STP or STL format
- Export of Surround View configuration for rapid prototyping
- Custom simulation scenes with pedestrians, vehicles, buildings
- Comprehensive PDF report generation
Intrinsic Calibration Tool
Automotive Cameras are complex systems, manufactured to highly precise tolerances. Still, no two cameras are exactly the same. The relative placement of lens and imager, lens decentration, thermal expansion and aging will cause two cameras mounted in the exact same location to produce slightly different images. Our intrinsic camera calibration tool can help determine and compensate for these inevitable build tolerances.
- Automated rectilinear, wide-angle and fisheye camera calibration
- Simple inexpensive setup yielding reprojection error values below 0.3
pixel (typical) for automotive lens
- Robust quality metric minimizing risk of operator error
- OpenCV-compatible output with Kannala model support
- Unique camera identification and tracking
- Check for long-term stability of camera parameters
- Automated plant-floor intrinsic quality verification according to ISO 17850:2015
Factory Calibration and Service Calibration Tool
Just like no two cameras are, neither are two cars. Camera mounting brackets, mirrors and other body components are subject to positioning tolerances. To perfectly stitch together images from cameras mounted at different locations factory calibration is typically used. Our factory / service calibration tool can utilize different types of targets to determine extrinsic camera parameters at subpixel accuracy.
Even a perfectly calibrated camera will slightly change its position on a vehicle over time, for example due to aging of mounting components. This can lead to misplacement of visual overlays, which no longer appear in the image exactly where they should.
Cogent has develop online calibration algorithms which can automatically compensate for these changes in camera position over time. Our algorithm works while the vehicle is moving (between 2 and 50 km/h). Calibration can be performed in less than 60 seconds with a stitching accuracy of 5cm @ 2m distance from the vehicle.
Cogent online calibration can be used for surround view systems (to support stitching accuracy), in rear camera and mirror replacement camera applications (e.g. to guarantee overlay position accuracy) and for lidar/camera co-calibration.
Our online calibration algorithm has been successfully implemented on various hardware platforms, including OpenGL, OpenCL, SSE4 / NEON based platforms, on R-Car IMPx5+ (R-Car V3M, V3H) and TI TDA2 (DSP C66x DSP).
UI and Mesh Configuration Tools
Our UI and Mesh Configuration Tools simplify various tasks encountered during development of surround view systems:
- UI layout, widgets, transitions, animations, screens configuration
- View configuration: 3D/2D, split, cross-traffic, single camera
- Integrated playback of Surround View from video files, network stream, static images
- Emulation of vehicle signals
- Optimal mesh density selection, rendering performance tuning
- Perspective distortion minimization
- Sizing of blending zones