METHOD AND APPARATUS FOR CALIBRATING ROLL ANGLE OF ON-BOARD CAMERA, AND DEVICE AND STORAGE MEDIUM

    公开(公告)号:EP4435723A1

    公开(公告)日:2024-09-25

    申请号:EP23777651.3

    申请日:2023-02-13

    发明人: WANG, Haichuan

    IPC分类号: G06T7/80

    摘要: The present application relates to the technical field of the Internet of Vehicles. Disclosed are a method and apparatus for calibrating a roll angle of an on-board camera, and a device and a storage medium. The embodiments of the present application can be applied to maps, autonomous driving, on-board or other scenarios. The method comprises: acquiring an image frame sequence, which is collected by an on-board camera during the traveling of a vehicle (510); selecting a plurality of image groups from the image frame sequence, wherein each image group comprises a plurality of consecutive image frames in the image frame sequence (520); acquiring lane line images, which respectively correspond to the plurality of image groups at a target roll angle, so as to obtain a lane line image set corresponding to the target roll angle (530); and according to lane line image sets respectively corresponding to different roll angles, determining a roll angle of the on-board camera (540).By means of the present application, a lane line error of the fitting of graphic frames which are obtained by means of photographing is reduced, thereby facilitating an improvement in the accuracy of roll angle calibration.

    OBJECT RECOGNIZING METHOD AND OBJECT RECOGNIZING DEVICE

    公开(公告)号:EP4390865A1

    公开(公告)日:2024-06-26

    申请号:EP22216476.6

    申请日:2022-12-23

    申请人: Moxa Inc.

    发明人: LEE, Wei-Yu

    摘要: The embodiments of the disclosure provide an object recognizing method and an object recognizing device (100). The method includes: obtaining (S210) a plurality of reference images (311-31N, P1, P2, 510, 520, 610, 620, 630), wherein the reference images (311-31N, P1, P2, 510, 520, 610, 620, 630) are captured by a camera, each reference image (311-31N, P1, P2, 510, 520, 610, 620, 630) includes a target object (33), and one of the camera and the target object has a moving speed higher than a first threshold; fusing (S220) the reference images (311-31N, P1, P2, 510, 520, 610, 620, 630) into a first fused image (320, P3, TS, 640); and recognizing (S230) the target object (33) corresponding to each reference image (311-31N, P1, P2, 510, 520, 610, 620, 630) via performing a first object detection on the first fused image (320, P3, TS, 640).