#StereoCamera #Telelian #CameraModule #ADAS #AI #2MP #3MP #8MP #HDR #CameraSensor
The Basic Process of Stereo Vision
Stereo vision is a method of acquiring images simultaneously with two cameras to extract depth information. The general procedure is as follows:
Image Acquisition: Capturing images at the exact same moment by synchronizing two cameras.
Camera Calibration: Extracting the intrinsic parameters of each camera and the extrinsic parameters between the two cameras.
Depth Map Generation: Calculating disparity using the aligned image pairs and estimating depth values.
Post-processing & Application: Correcting the depth map and applying it to various fields such as point cloud generation, object recognition, and robot navigation.
While this process is algorithmically well-established, camera selection significantly impacts performance in actual implementation.
Limitations of General Stereo Cameras
Standard USB-based stereo cameras perform reasonably well in indoor settings. However, they face significant issues in outdoor environments. Image quality tends to drop drastically, particularly under direct sunlight or in backlit conditions.
Furthermore, since most utilize USB-C interfaces, various hardware issues arise, including power supply instability, cable length limitations, and connection failures. Many posts on developer forums complain about these specific issues, and we experienced the same difficulties during actual usage.
Strengths of ADAS Cameras
Automotive ADAS cameras are a product line already proven in diverse and harsh environments. They demonstrate stable performance under backlight, strong sunlight, and rapid changes in lighting conditions. Looking at Telelian's existing camera lineup, it is easy to understand why these cameras are widely used in vehicles and outdoor robots.
Development of the GMSL-based Stereo Camera
Based on these limitations and our experiences, we proceeded to develop a stereo camera utilizing GMSL ADAS cameras. Initially conceived as a simple idea, the project was officially launched due to a specific need for a stereo camera capable of operating in outdoor environments for an actual project.
The biggest differentiator during the development process was hardware synchronization and the GMSL interface. We were able to precisely synchronize two cameras using external triggers and simultaneously resolve issues related to cable length and power supply.
Results and Significance
Consequently, we secured performance that is much more stable and reliable compared to existing USB-based stereo cameras. The results exceeded our expectations, and the system has been immediately deployed in active projects.
We reached the conclusion that for outdoor robots, autonomous driving systems, and ADAS research, utilizing proven automotive ADAS cameras is far more effective than relying on simple consumer-grade cameras.
Conclusion
Algorithms alone are not sufficient for stereo vision; selecting the right camera for the environment is essential. While general stereo cameras may be appropriate for indoors, ADAS camera-based solutions are far more suitable for outdoor use. We expect stereo systems utilizing GMSL ADAS cameras to play a pivotal role in future outdoor robots and autonomous platforms.
#StereoCamera #Telelian #CameraModule #ADAS #AI #2MP #3MP #8MP #HDR #CameraSensor
The Basic Process of Stereo Vision
Stereo vision is a method of acquiring images simultaneously with two cameras to extract depth information. The general procedure is as follows:
Image Acquisition: Capturing images at the exact same moment by synchronizing two cameras.
Camera Calibration: Extracting the intrinsic parameters of each camera and the extrinsic parameters between the two cameras.
Depth Map Generation: Calculating disparity using the aligned image pairs and estimating depth values.
Post-processing & Application: Correcting the depth map and applying it to various fields such as point cloud generation, object recognition, and robot navigation.
While this process is algorithmically well-established, camera selection significantly impacts performance in actual implementation.
Limitations of General Stereo Cameras
Standard USB-based stereo cameras perform reasonably well in indoor settings. However, they face significant issues in outdoor environments. Image quality tends to drop drastically, particularly under direct sunlight or in backlit conditions.
Furthermore, since most utilize USB-C interfaces, various hardware issues arise, including power supply instability, cable length limitations, and connection failures. Many posts on developer forums complain about these specific issues, and we experienced the same difficulties during actual usage.
Strengths of ADAS Cameras
Automotive ADAS cameras are a product line already proven in diverse and harsh environments. They demonstrate stable performance under backlight, strong sunlight, and rapid changes in lighting conditions. Looking at Telelian's existing camera lineup, it is easy to understand why these cameras are widely used in vehicles and outdoor robots.
Development of the GMSL-based Stereo Camera
Based on these limitations and our experiences, we proceeded to develop a stereo camera utilizing GMSL ADAS cameras. Initially conceived as a simple idea, the project was officially launched due to a specific need for a stereo camera capable of operating in outdoor environments for an actual project.
The biggest differentiator during the development process was hardware synchronization and the GMSL interface. We were able to precisely synchronize two cameras using external triggers and simultaneously resolve issues related to cable length and power supply.
Results and Significance
Consequently, we secured performance that is much more stable and reliable compared to existing USB-based stereo cameras. The results exceeded our expectations, and the system has been immediately deployed in active projects.
We reached the conclusion that for outdoor robots, autonomous driving systems, and ADAS research, utilizing proven automotive ADAS cameras is far more effective than relying on simple consumer-grade cameras.
Conclusion
Algorithms alone are not sufficient for stereo vision; selecting the right camera for the environment is essential. While general stereo cameras may be appropriate for indoors, ADAS camera-based solutions are far more suitable for outdoor use. We expect stereo systems utilizing GMSL ADAS cameras to play a pivotal role in future outdoor robots and autonomous platforms.