Methods for comparing a target block to a reference window for motion estimation during video encoding

• US 9,497,482 B2
• Filed: 10/02/2013
• Issued: 11/15/2016
• Est. Priority Date: 10/02/2013
• Status: Active Grant

• Current Assignee: Amlogic (Shanghai) Co., Ltd. (Amlogic Holdings Ltd.)
• Original Assignee: Amlogic (Shanghai) Co., Ltd. (Amlogic Holdings Ltd.)
• Inventors: Chen, Shi
• Primary Examiner(s): Dastouri, Mehrdad
• Assistant Examiner(s): Walker, Jared
• Application Number:
  US14/044,739
• Publication Number:
  US 20150092835A1
• Time in Patent Office: 1,140 Days
• Field of Search: 37524001-24029
• US Class Current: 1/1
• CPC Class Codes: H04N 19/43 Hardware specially adapted for motion estimation or compensation
• CPC Class Codes: H04N 19/53 Multi-resolution motion estimation; Hierarchical motion estimation

Abstract

A method for comparing a target block to a reference window in motion estimation calculations, comprises the steps of: determining SAD calculations as a function of the target block and the reference window; reading the target block into registers; reading a segment of the reference window; updating the determined SAD calculations on the fly as a function of the read target block and the read segment of the reference window; and determining one or more sub-blocks of the reference window having minimum SAD values as a function of the updated SAD calculations; and if one or more of the minimum SAD values are not found, repeating the reading a segment step, the updating step and the determining step for a next segment of the reference window.

Claims

1. A method for comparing a target block to a reference window in motion estimation calculations, comprising the steps of:

determining SAD calculations as a function of the target block and the reference window, wherein the determined SAD calculations are calculated by summing the absolute differences of pixel data of the target block and pixel data of the reference window;

reading the target block into registers;

reading a segment of the reference window, wherein the reference is an array of pixels, wherein each row of the array of pixels is partitioned into a left segment, a middle segment, and a right segment, and wherein an order that the segments are read by a processor is as follows;

the middle segment, the left segment, and the right segment;

updating the determined SAD calculations on the fly as a function of the read target block and the read segment of the reference window;

determining one or more sub-blocks of the reference window having minimum SAD values as a function of the updated SAD calculations; and

if one or more of the minimum SAD values are not below a perdefined threshold, repeating the reading a segment step, the updating step and the determining step for a next segment of the reference window.

2. The method of claim 1 wherein if a certain one of the updated SAD calculations is completed for a certain sub-block of the reference window and is below a predefined threshold, using the certain sub-block of the reference window for motion estimation.

3. The method of claim 1 wherein if the updated SAD calculations are completed and are equal to or above a predefined threshold, using the determined one or more sub-blocks of the reference window for motion estimation.

4. The method of claim 1 wherein the reference window is an array of pixels and wherein, in the reading the segments step, a middle row of the array of pixels is read first and then alternating top and bottoms rows, relative to the middle row, are read in order of proximity to the middle row.

5. The method of claim 1 wherein the target block has 4 rows by 4 columns of pixels and wherein the target block is a down sampled block from a target image.

6. The method of claim 1 wherein the reference window has 20 rows by 20 columns of pixels and wherein the reference window is a down-sampled block from a reference image.

7. The method of claim 1 wherein the determined SAD calculations are weighted as a function of a previously determined sub-block having a minimum SAD calculation.

8. The method of claim 1 wherein the determined SAD calculations are weighted based upon a skip flag for any of the sub-blocks of the reference window.

9. A method for comparing a target block to a reference window in motion estimation calculations, comprising the steps of:

determining SAD calculations as a function of pixel data of the target block and pixel data of the reference window;

reading the target block into registers;

reading a segment of the reference window, wherein the reference window is an array of pixels, and wherein a middle row of the array of pixels is read first and then alternating top and bottoms rows, relative to the middle row, are read in order of proximity to the middle row;

updating the determined SAD calculations on the fly as a function of the read target block and the read segment of the reference window;

determining one or more sub-blocks of the reference window having minimum SAD values as a function of the updated SAD calculations, wherein if a certain one of the updated SAD calculations is completed for a certain sub-block of the reference window and is below a predefined threshold, using the certain sub-block of the reference window for motion estimation; and

if one or more of the minimum SAD values are not below the predefined threshold, repeating the reading a segment step, the updating step and the determining step for a next segment of the reference window,

wherein if the updated SAD calculations are completed and are equal to or above the predefined threshold, using the determined one or more sub-blocks of the reference window for motion estimation.

10. The method of claim 9 wherein the reference window is an array of pixels, wherein each row of the array of pixels is partitioned into a first segment, a second segment, and a third segment, and wherein, in the reading the segments step, the order that the segments are read by the processor is as follows:

the second segment, the first segment, and the third segment.

11. The method of claim 9 wherein the target block has 4 rows by 4 columns of pixels and wherein the target block is a down sampled block from a target image.

12. The method of claim 9 wherein the reference window has 20 rows by 20 columns of pixels and wherein the reference window is a down-sampled block from a reference image.

13. The method of claim 9 wherein the determined SAD calculations are weighted as a function of a previously determined sub-block having a minimum SAD calculation.

14. The method of claim 9 wherein the determined SAD calculations are weighted based upon a skip flag for any of the sub-blocks of the reference window.

15. A method for comparing a target block to a reference window in motion estimation calculations, comprising the steps of:

determining SAD calculations as a function of the target block and the reference window;

reading the target block into registers;

reading a segment of the reference window;

updating the determined SAD calculations on the fly as a function of the read target block and the read segment of the reference window;

determining one or more sub-blocks of the reference window having minimum SAD values as a function of the updated SAD calculations, wherein if a certain one of the updated SAD calculations is completed for a certain sub-block of the reference window and is below a predefined threshold, using the certain sub-block of the reference window for motion estimation; and

if one or more of the minimum SAD values are not below the predefined threshold, repeating the reading a segment step, the updating step and the determining step for a next segment of the reference window,

wherein if the updated SAD calculations are completed and are equal to or above the predefined threshold, using the determined one or more sub-blocks of the reference window for motion estimation,

wherein the reference window is an array of pixels,

wherein each row of the array of pixels is partitioned into a first segment, a second segment, and a third segment,

wherein, in the reading the segments step, the order that the segments are read by the processor is as follows;

the second segment, the first segment, and the third segment, and

wherein, in the reading the segments step, the segments in a middle row of the array of pixels is read first and then the alternating top and bottoms rows, relative to the middle row, are read in order of proximity to the middle row.

16. The method of claim 15 wherein the target block has 4 rows by 4 columns of pixels, wherein the target block is a down sampled block from a target image, wherein the reference window has 20 rows by 20 columns of pixels and wherein the reference window is a down-sampled block from a reference image.

17. The method of claim 15 wherein the determined SAD calculations are weighted as a function of a previously determined sub-block having a minimum SAD calculation.

18. The method of claim 15 wherein the determined SAD calculations are weighted based upon a skip flag for any of the sub-blocks of the reference window.