System and method for estimating motion vector in macro block

• US 6,031,582 A
• Filed: 12/19/1997
• Issued: 02/29/2000
• Est. Priority Date: 12/20/1996
• Status: Expired due to Term

• Current Assignee: Kabushiki Kaisha Toshiba (Toshiba Corporation)
• Original Assignee: Kabushiki Kaisha Toshiba (Toshiba Corporation)
• Inventors: Tsuboi, Yoshiro, Takahashi, Masafumi, Nishikawa, Tsuyoshi
• Primary Examiner(s): Le, Vu
• Application Number:
  US08/993,747
• Time in Patent Office: 802 Days
• Field of Search: 348/402, 348/407, 348/413, 348/416, 348/699, 386/111, 382/236
• US Class Current: 348/699
• CPC Class Codes: H04N 19/51 Motion estimation or motion compensation
• CPC Class Codes: H04N 19/557 Motion estimation characterised by stopping computation or iteration based on certain criteria, e.g. error magnitude being too large or early exit

Abstract

A motion vector estimating system prepares a predetermined range of a frame/field of image data as a macro block, selects a plurality of proposed motion vectors in the reference region, calculates a signal-strength difference between pixels corresponding to each of the proposed motion vectors of current and reference images. The system comprises: an accumulating element for deriving an accumulation result in parallel on the basis of the signal-strength difference by means of accumulating circuits connected in parallel as a pipeline; an accumulation stopping element having separate stop circuits for separately stopping the accumulation of the accumulating element every n accumulating circuits; and a control element for controlling the supply of the current image data and the reference image data for proposed motion vectors for each of the accumulating circuits, and for outputting a separate stop signal to the accumulation stopping element for stopping the accumulation for the proposed motion vector of the accumulating circuits immediately after the accumulated value of the absolute values of the differences between the signal values of picture elements of the current image and the reference image exceeds a predetermined value. Thus, it is possible to suitably omit unnecessary accumulation in a pipeline processing for estimating a motion vector, so that it is possible to increase the processing speed and to decrease the electric power consumption.

Claims

1. A system for estimating a motion vector for a macro block, which designates an image in a predetermined range of a frame/field in an image data as a macro block to select proposed motion vectors in the macro block, calculates a signal-strength difference between picture elements corresponding to the proposed motion vectors of a current image and a reference image, accumulates the signal-strength differences calculated for the picture elements in the macro block to derive an accumulation result, and selects a proposed motion vector providing an accumulation result of a maximum or minimum value from the proposed motion vectors in a specific range of the macro block as a motion vector, said system comprising:

accumulating means comprising a plurality of accumulating circuits, connected as a pipeline, for inputting a current image data and a reference image data in parallel for the proposed motion vectors and for parallel-processing accumulation for a plurality of motion vectors;

accumulation stopping means for separately stopping an accumulation processing operation of the accumulating means for each of the accumulating circuits; and

control means for controlling the supply of the current image data and the reference image data for the proposed motion vectors for each of the accumulating circuits of the accumulating means, and for outputting a separate stop signal for stopping a separate accumulation for each of the accumulating circuits to control the stop of the accumulation processing of the accumulation circuit.

2. The motion vector estimating system as set forth in claim 1,

wherein each of said accumulating circuits of said accumulating means has a determining circuit having a predetermined identification threshold for comparing a partway accumulated value with the identification threshold at least once to determine whether an accumulated value is valid or invalid, on the basis of the comparison result, and

wherein said stopping means has a separate stop circuit for stopping a corresponding accumulating circuit until the accumulation operation for a newly proposed motion vector is started, when the determination result of said determining circuit is invalid.

3. The motion vector estimating system as set forth in claim 2, wherein said accumulating means has a start circuit for disqualifying all of the proposed motion vectors currently reviewed to start the accumulation processing operation for a newly proposed motion vector, immediately after all of the accumulated values compared with the identification threshold in the parallel processing are determined to be invalid.

4. The motion vector estimating system as set forth in claim 3, wherein said stopping means has a stop circuit for outputting a motion vector estimation terminating signal when the motion vector estimation is completed, to stop the operation of said accumulating means until a motion vector estimation request signal is inputted from the outside.

5. The motion vector estimating system as set forth in claim 4, wherein said accumulating means determines said identification threshold on the basis of the maximum or minimum value of the accumulated values which have been reviewed.

6. The motion vector estimating system as set forth in claim 4, wherein said accumulating means determines said identification threshold using a fixed value, which does not change during a motion vector estimating process, on the basis of a predetermined rate to a theoretically available maximum value.

7. The motion vector estimating system as set forth in claim 6, wherein each of said accumulating circuits of said accumulating means has a flip-flop capable of retaining a maximum number which is equal to said identification threshold, and deals with a carry-out signal of the most significant digit of each of the accumulating circuits as a magnitude comparing signal compared with said identification threshold so as to decrease the number of digits of the flip-flop for retaining the accumulated values.

8. A method for estimating a motion vector for a macro block in a reference region, said method comprising the steps of:

designating an image in a predetermined range of a frame/field as a macro block, and selecting a plurality of proposed motion vectors existing in the reference region;

calculating, in parallel, signal-strength differences between corresponding picture elements of a current image and a reference image for a predetermined parallel number of proposed motion vectors of said plurality of proposed motion vectors in said macro block designated, and accumulating said signal-strength differences for each of specific proposed motion vectors processed in parallel; and

stopping the accumulation for the proposed motion vectors when the accumulation result of the minimum or maximum value is identified for any one of the parallel number of proposed motion vectors processed in parallel.

9. The motion vector estimating method as set forth in claim 8, wherein said step of stopping the accumulation comprises the steps of:

comparing the magnitudes of a partway accumulated value for each of said proposed motion vectors processed in parallel and a predetermined identification threshold at least once;

determining whether the accumulated value for each of said proposed motion vectors is valid or invalid, on the basis of the compared result; and

stopping the accumulation operation for the corresponding one of said proposed motion vectors until the accumulation for a newly proposed motion vector is started, when it is determined that the accumulated value is invalid.

10. The motion vector estimating method as set forth in claim 8, which, after said step of stopping accumulation, further comprises the step of:

disqualifying all of said proposed motion vectors, for each of which the accumulation is carried out by the parallel processing, and starting a parallel processing for a newly proposed motion vector, immediately after all of the accumulation results for all of said proposed motion vectors processed in parallel are stopped.