利索-全球专利检索

  1. 登录
  2. 注册

搜索结果

34 条记录 (耗时 0.01 秒)

    SYSTEMS AND METHODS FOR HYBRID DEPTH REGULARIZATION
    10.
    发明公开
    SYSTEMS AND METHODS FOR HYBRID DEPTH REGULARIZATION 审中-公开

    公开(公告)号：EP4432231A2

    公开(公告)日：2024-09-18

    申请号：EP24171265.2

    申请日：2018-08-17

    申请人： Adeia Imaging LLC

    发明人： Jain, Ankit K. ,  Chatterjee, Priyam ,  Venkataraman, Kartik

    IPC分类号： G06T7/593

    CPC分类号： G06T5/30 ,  G06T5/50 ,  G06T2207/2002820130101 ,  G06T2207/2003220130101 ,  G06T2207/2003620130101 ,  G06T2207/2019220130101 ,  G06T7/136 ,  G06T7/593 ,  G06T7/13 ,  G06T2207/1001220130101

    摘要： A computer-implemented depth sensing method, comprising: obtaining (302) image data for a plurality of images from multiple viewpoints, wherein the image data for the plurality of images comprises a reference image (400) and at least one alternate view image; generating a raw depth map (410) containing depth estimates for pixels within the reference image (400) using the image data for the reference image (400) and the image data for the at least one alternate view image using a first depth estimation process, and a confidence map (420) describing reliability of depth estimates contained within the raw depth map (410), wherein generating the raw depth map (410) and the confidence map (420) further comprises: measuring parallax observable between the reference image (400) and the at least one alternate view image by comparing the similarity of a pixel in the reference image (400) to pixels in the at least one alternate view image determined based upon a plurality of depth samples using a cost function; and estimating, based upon the measured parallax, depth for the pixel in the reference image (400) by identifying the sampled depth at which the cost function for a pixel in the reference image (400) indicates the strongest match as being the estimated depth of the pixel; generating a regularized depth map (480) by: computing a secondary depth map (430) containing depth estimates for pixels within the reference image (400) using a second depth estimation process different from the first depth estimation process by: generating cost volume using costs determined using the cost function at each sampled depth; and computing the secondary depth map (430) based on the cost volume; and computing a first composite depth map (440) by selecting depth estimates from the raw depth map (410) and the secondary depth map (430), where a depth estimate for a pixel in the reference image (400) is selected from the raw depth map (410) when the depth estimate is indicated as being reliable by the confidence map (420).