Depth mask assisted video stabilization
    11.
    发明授权
    Depth mask assisted video stabilization 有权
    深度掩模辅助视频稳定

    公开(公告)号:US09100574B2

    公开(公告)日:2015-08-04

    申请号:US14346390

    申请日:2011-10-18

    IPC分类号: H04N5/232 G06T5/00

    摘要: Systems (490), methods (340), and machine-readable and executable instructions (428) are provided for depth mask assisted video stabilization. Depth mask assisted video stabilization can include creating a first depth mask from a first frame (101, 201) and a second depth mask from a second frame (102, 202) to obtain depth information (342) and defining a first list of feature points (111-1a, 111-2a, 111-3a, 111-4a, 113-1a, 113-2a, 113-3a, 113-4a, 113-5a) from the first frame (101, 201) and a second list of feature points (111-1b, 111-2b, 111-3b, 111-4b, 113-1b, 113-2b, 113-3b, 113-4b, 113-5b) from the second frame (102, 202) (344). Depth mask assisted video stabilization can also include comparing feature points from the first list of feature points (111-1a-111-4a, 113-1a-113-5a) with feature points from the second list of feature points (111-1b-111-4b, 113-1b-113-5b) to find a number of common feature point pairs (111-1a, 111-1b), (111-2a, 111-2b), (111-3a, 111-3b), (111-4a, 111-4b), (113-1a, 113-1b), (113-2a, 113-2b), (113-3a, 113-3b), (113-4a, 113-4b), (113-5a, 113-5b)) (345) and creating a number of motion matrices from the common feature point pairs ((111-1a, 111-1b), (111-2a, 111-2b), (111-3a, 111-3b), (111-4a, 111-4b), (113-1a, 113-1b), (113-2a, 113-2b), (113-3a, 113-3b), (113-4a, 113-4b), (113-5a, 113-5b)) (346). Depth mask assisted video stabilization can also include including the motion matrices in a correction matrix (348) and applying the inverse of the correction matrix to the second frame (102, 202) to perform video stabilization (350).

    摘要翻译: 系统(490),方法(340)和机器可读和可执行指令(428)被提供用于深度掩模辅助视频稳定。 深度掩模辅助视频稳定可以包括从第一帧(101,201)创建第一深度掩码和从第二帧(102,202)创建第二深度掩码以获得深度信息(342)并且定义特征点的第一列表 (101,201)的第一列(111-1a,111-2a,111-3a,111-4a,113-1a,113-2a,113-3a,113-4a,113-5a) 的特征点(111-1b,111-2b,111-3b,111-4b,113-1b,113-2b,113-3b,113-4b,113-5b)从第二帧(102,202)( 344)。 深度掩模辅助视频稳定还可以包括将来自特征点(111-1a-111-4a,113-1a-113-5a)的第一列表的特征点与来自第二特征点列表(111-1b- (111-1a,111-1b),(111-2a,111-2b),(111-3a,111-3b),(111-1a,111-2b), ,(111-4a,111-4b),(113-1a,113-1b),(113-2a,113-2b),(113-3a,113-3b),(113-4a,113-4b) (111-1a,111-1b),(111-2a,111-2b),(111),(113-5a,113-5b))(345) (113-1a,111-3b),(111-4a,111-4b),(113-1a,113-1b),(113-2a,113-2b),(113-3a,113-3b),(113 -4a,113-4b),(113-5a,113-5b))(346)。 深度掩模辅助视频稳定还可以包括在校正矩阵(348)中包括运动矩阵,并将校正矩阵的逆应用于第二帧(102,202)以执行视频稳定(350)。