公开(公告)号：JP2000194826A

公开(公告)日：2000-07-14

申请号：JP31627899

申请日：1999-11-08

Applicant: IBM

Inventor： CHAPMAN SYDNEY GEORGE ,  GALTON BRIAN NEIL ,  SHARP CHRISTOPHER EDWARD

IPC: G06T3/00 ,  H04N5/262 ,  H04N5/33 ,  G06T1/00

Abstract: PROBLEM TO BE SOLVED: To relatively position images to each other and to identify an appropriate connection line by positioning plural signal transmitters within the visual field overlapping area of plural cameras and making them tie points in the images. SOLUTION: Plural infrared transmitters 30 generate tie points to identify a connection line when a pair of cameras 10 and 20 having visual fields which are different but overlap on each other fetches images and connects them with an appropriate line. In such a case, each camera unit has a beam splitter 50 that makes light from the same viewpoint of a scene direct to both a camera 60 responding to visible spectrum light and an infrared camera 70 to respond to a signal from the transmitters 30 and to identify a tie point, and the tie point is visible in an image from the camera 70. A filter in an image processing component 80 eliminates the whole infrared radiation except the tie point according to the minimum intensity threshold. Thus, the tie point is outputted as image coordinates of a bright point.

公开(公告)号：GB2345215A

公开(公告)日：2000-06-28

申请号：GB9828141

申请日：1998-12-22

Applicant: IBM

Inventor： CHAPMAN SYDNEY GEORGE ,  GALTON BRIAN NEIL ,  SHARP CHRISTOPHER EDWARD

IPC: G06T3/00 ,  H04N5/262 ,  H04N5/33 ,  H04N5/222

Abstract: Tie-points for use in image registration are automatically identified, such as when a final image is desired to have a wider field of view than is obtainable by conventional cameras 10,20. A plurality of infra-red signal transmitters 30 are positioned within a scene, each acting as a self-identifying tie-point when captured on an infra-red camera. The coordinates of tie-points identified from the infra-red image are then associated with images captured by visible image cameras and can be used by image warping software for combining the visible images.

公开(公告)号：GB2327822B

公开(公告)日：2002-02-27

申请号：GB9715649

申请日：1997-07-25

Applicant: IBM

Inventor： GALTON BRIAN NEIL

IPC: H04N7/26 ,  H04N7/30 ,  H04N7/46

公开(公告)号：GB2345215B

公开(公告)日：2003-06-25

申请号：GB9828141

申请日：1998-12-22

Applicant: IBM

Inventor： CHAPMAN SYDNEY GEORGE ,  GALTON BRIAN NEIL ,  SHARP CHRISTOPHER EDWARD

IPC: G06T3/00 ,  H04N5/262 ,  H04N5/33 ,  H04N5/222

公开(公告)号：GB2327822A

公开(公告)日：1999-02-03

申请号：GB9715649

申请日：1997-07-25

Applicant: IBM

Inventor： GALTON BRIAN NEIL

IPC: H04N7/26 ,  H04N7/30 ,  H04N7/46

Abstract: A method for processing digitised video material comprising the steps of: capturing 4 a video sequence having a first start-time and a first end-time to provide plurality of digitized frames; filtering 6-13 the frames according to a pre-determined filter function; decimating 6-13 the frames to reduce the frames to a predetermined horizontal and vertical frame size; and encoding 15 the filtered and decimated frames. In response to the encoding step failing to encode the data at a failure time between the first start-time and the first end-time, the video sequence is divided 20 into two or more sequences having a respective start-time and end-time. One of the sequences, having a start-time and an end-time temporally including the failure time, is selected; and the filter function for the selected sequence is re-defined 21. The capturing, filtering, decimating and encoding steps are then repeated for the selected sequence.