Camera-LiDAR joint calibration is the basic requirement for data fusion of two sensors in autonomous vehicle. In order to minimize the re-projection error of the joint calibration and enhance the accuracy of the fusion of image and laser point cloud data, a Levenberg-Marquardt algorithm is designed to optimize the external parameters for the joint calibration of the camera and LiDAR. Firstly, the least square function is constructed based on the difference between image pixel coordinates and point cloud re-projection coordinates. To reduce the number of optimization parameters, we employ the Rodriges transformation, a formula that enables us to represent the rotation matrix using a vector of only 3 parameters. Then, the Levenberg-Marquardt algorithm is utilized to optimize the joint calibration parameters and solve the optimized re-projection error. The results demonstrate that the optimized re-projection error is 0.226, which is reduced by 53.5% compared with before optimization, and the accuracy of joint calibration is improved. Finally, visualization of the calibration results validates the effectiveness of the optimization algorithm.
Due to the frequent occurrence of pedestrian accidents caused by the blind spot of commercial vehicle drivers, the YOLOv5s-FE algorithm is proposed to detect pedestrians in the blind spot using YOLOv5s as a baseline model for this problem. The Swin Transformer Block structure is introduced in the Neck network to improve the feature extraction capability to capture the global information. To solve the problem of missed detection of small-target pedestrians, the feature fusion part in the network structure is added with an upsampling layer to get a larger size feature map and then fused with shallow features to preserve the localization information of small-target pedestrians. Through experimental verification, the precision, recall and accuracy are improved by 2.2%, 2.1%, and 3.4%, respectively, and the pedestrian target can be detected effectively.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.