Camera work detecting method

• US 5,267,034 A
• Filed: 02/25/1992
• Issued: 11/30/1993
• Est. Priority Date: 03/11/1991
• Status: Expired due to Fees

• Current Assignee: Director-General of the Agency of Industrial Science and Technology, the Ministry of International Trade and Industry of Japan (Government of Japan)
• Original Assignee: Institute For Personalized Information Environment
• Inventors: Miyatake, Takafumi, Ueda, Hirotada, Yoshizawa, Satoshi
• Primary Examiner(s): Britton, Howard W.
• Application Number:
  US07/840,825
• Time in Patent Office: 644 Days
• Field of Search: 358/105, 358/136, 358/222
• US Class Current: 348/352
• CPC Class Codes: G11B 2220/20 Disc-shaped record carriers
• CPC Class Codes: G11B 2220/90 Tape-like record carriers
• CPC Class Codes: G11B 27/11 by using information not detectable on the record carrier
• CPC Class Codes: G11B 27/28 by using information signals recorded by the same method as the main recording G11B27/22 takes precedence
• CPC Class Codes: H04N 23/68 for stable pick-up of the scene, e.g. compensating for camera body vibrations
• CPC Class Codes: H04N 23/6812 based on additional sensors, e.g. acceleration sensors
• CPC Class Codes: H04N 23/6815 by distinguishing pan or tilt from motion

Abstract

A motion picture image editing method and in particular, a method for detecting such camera works as zooming and panning motions from a motion picture image stored in video tape or the like. In this method, a motion picture image constituted by a plurality of consecutive picture images on every frame basis, and on time series basis and on the basis of a correlation value for a displacement between the frames at a typical point determined for each of small blocks constituting the motion picture image, a motion vector for each of the small blocks is detected. Then a first motion parameter for estimation of a motion of a camera is generated with use of the motion vector detected at corresponding one of the typical points of first one of the frames. Or the first motion parameter for estimation of the motion of the camera is generated with use of a combination of the motion vector and a position vector at corresponding one of the typical points of a second frame inputted previous by one frame to the first frame. Further, the first motion parameters for the respective frames are found, the found first motion parameters are integrated until an integrated value exceeds a predetermined allowable value, and the integrated value is averaged through an integration period to calculate a second motion parameter for determination of the camera motion for the integration period. Finally, the camera motion is detected on the basis of the second motion parameter.

Claims

1. A camera work detecting method comprising the steps of:

inputting a motion picture image constituted by a plurality of consecutive picture images on every frame basis and on time series basis and on the basis of a correlation value for a displacement between said frames at a typical point determined for each of small blocks constituting said motion picture image, detecting a motion vector for each of said small blocks;

generating a first motion parameter for estimation of a motion of a camera with use of either one of the motion vector detected at corresponding one of said typical points of first one of said frames and a combination of said motion vector and a position vector at corresponding one of the typical points of a second frame inputted previous by one frame to said first frame;

finding said first motion parameters for said respective frames and integrating the found first motion parameters until an integrated value exceeds a predetermined allowable value;

averaging said integrated value through an integration period and calculating a second motion parameter for determination of said camera motion for the integration period; and

detecting said camera motion on the basis of said second motion parameter.

2. A camera motion detecting method as set forth in claim 1, wherein one of said small blocks of the picture image of said second frame having a non-uniform image pattern is used as a standard dictionary, a pattern coinciding with the pattern of said standard dictionary is found in the vicinity of one of said small blocks in said first frame corresponding to the standard dictionary, and said motion vector is detected on the basis of a difference between a position of said found pattern in said first frame and a position of said standard dictionary in said second frame.

3. A camera motion detecting method as set forth in claim 2, wherein said difference between the position of said found pattern in said first frame and the position of said standard dictionary in said second frame is found on the basis of a given position in the pattern of said standard dictionary as a reference point.

4. A camera work detecting method as set forth in claim 1, wherein said first motion parameter for estimation of the motion of said camera is found with use of the motion vector detected at corresponding one of said typical points of first one of said frames by finding statistical deviations of directions and magnitudes of the motion vectors at the respective typical points determined for the respective small blocks, by calculating an average of said typical point motion vectors having the deviations within a predetermined allowable range and by finding a motion parameter of said first motion parameters associated with rotation motion around a horizontal or vertical axis of said camera with use of said average motion vector;

and

said first motion parameter for estimation of the motion of said camera is found with use of said combination of said motion vector and the position vector at corresponding one of the typical points of said second frame inputted previous by one frame to said first frame, by extracting ones of the motion vectors having directions included in a predetermined allowable range in which a directional difference between each motion vector and the associated position vector at the typical point of each small block is within said predetermined allowable range, by calculating inner products of said extracted motion vectors and unit vectors of said associated position vectors, by normalizing values corresponding to additions of said inner products and magnitudes of said position vectors with use of the magnitudes of said position vectors to find photographing magnifications, by finding a statistical deviation of said found photographing magnifications, by calculating an average of ones of the photographing magnifications having the deviations within a predetermined allowable range, and by finding parameters indicative of a lens motion of said camera with use of said average photographing magnification.

5. A camera work detecting method as set forth in claim 4, wherein said second motion parameter based on said first motion parameter indicative of a rotation motion of said camera as well as said second motion parameter based on said first motion parameter indicative of a lens motion of said camera are respectively separately registered in different tables with respect to different frames, and the motion of said camera is detected on every frame basis by retrieving said tables.