IP Library Granted Patent US 12,737,894
Granted Patent B2
US 12,737,894 · App. 18/521,823 · Granted Sep 15, 2026

Methods and systems for motion vector filtering

Inventors: Guohua Cheng (Shanghai, CN); Neil Woodall (Newport Beach, CA); Hongmin Zhang (Santa Clara, CA)
Assignee: PIXELWORKS SEMICONDUCTOR TECHNOLOGY (SHANGHAI) CO. LTD.
G06T7/20G06T5/77G06T2207/20024
View Patent ↗
Loading inventors, assignments & file history…
Monitor This Case
Get email alerts when status or documents change.
Order Certified Copies
Most orders are placed with the USPTO same day — all within 24 business hours.
Order via The Patent Place →
Pre-filled with this patent's details
Quick Facts
Patent No.
US 12,737,894
App. No.
18/521,823
Granted
Sep 15, 2026
Kind
B2
Abstract

Systems and methods are herein provided for motion vector filtering to reduce hole size. In one example, a method comprises determining presence of one or more holes in an outputted image, wherein the outputted image was generated based on one or more motion vector (MV) fields and depth fields; determining a local foreground MV field and depth field for each pixel of the outputted image; defining a hole MV field based on the local foreground MV field; filter the hole MV field along its motion trajectory to reduce size of the one or more holes; and outputting a filtered image with one or more smaller holes.

Claims (13)

1 . A method, comprising:

determining presence of one or more holes in an outputted image, wherein the outputted image was generated based on one or more motion vector (MV) fields and depth fields;

determining a local foreground MV field and depth field for each pixel of the outputted image;

defining a hole MV field based on the local foreground MV field;

filter the hole MV field along its motion trajectory to reduce size of the one or more holes; and

outputting a filtered image with one or more smaller holes.

2 . The method of claim 1 , further comprising performing an inpainting algorithm on the filtered image to fill the one or more smaller holes.

3 . The method of claim 1 , wherein filtering the hole MV field along its motion trajectory comprises determining an MV at a projected position for each step along the motion trajectory.

4 . The method of claim 3 , wherein filtering the hole MV field further comprises comparing an MV of a given pixel and the MV at the projected position of that pixel in the outputted image to determine which MVs to filter.

5 . The method of claim 4 , further comprising determining foreground and background relationship between the MV of the given pixel and the MV at the projected position.

6 . The method of claim 1 , wherein filtering the hole MV field along its motion trajectory comprises accumulating MVs within a specified range to determine a filtered MV field.

7 . The method of claim 6 , wherein the filtered MV field comprises the one or more smaller holes.

8 . The method of claim 6 , further comprising performing foreground MV dilation on the filtered MV field.

Assignments (2)
CHANGE OF ASSIGNEE ADDRESS Recorded Mar 19, 2026
From: PIXELWORKS SEMICONDUCTOR TECHNOLOGY (SHANGHAI) CO. LTD.
To: PIXELWORKS SEMICONDUCTOR TECHNOLOGY (SHANGHAI) CO. LTD.
Reel/Frame 075146/0897 →
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded Nov 28, 2023
From: CHENG, GUOHUA; WOODALL, NEIL; ZHANG, HONGMIN
To: PIXELWORKS SEMICONDUCTOR TECHNOLOGY (SHANGHAI) CO. LTD.
Reel/Frame 065689/0107 →
Continuity (1)
Related Publication 20250173875A1 · May 29, 2025
References Cited (19)
US 10142651B1 · Cheng et al. · 2018 [cited by applicant]
US 20110142289A1 · Barenbrug et al. · 2011 [cited by applicant]
US 20110216831A1 · Rossignol · 2011 [cited by examiner]
US 20210398291A1 · Sasaki · 2021 [cited by examiner]
US 20230115057A1 · Zhang et al. · 2023 [cited by applicant]
US 20240214607A1 · Li · 2024 [cited by examiner]
US 20240257358A1 · Kimura · 2024 [cited by examiner]
US 20250173877A1 · Cheng · 2025 [cited by examiner]
CN 106954076A · 2017 [cited by applicant]
WO 2023150488A1 · 2023 [cited by applicant]
Criminisi, A. et al., “Region filling and object removal by exemplar-based image inpainting,” IEEE Transactions on Image Processing, vol. 13, No. 9, Sep. 2004, Available Online Aug. 16, 2004, 13 pages. [cited by applicant]
Xu, Z. et al., “Image Inpainting by Patch Propagation Using Patch Sparsity,” IEEE Transactions on Image Processing, vol. 19, No. 5, May 2010, Available Online Feb. 2, 2010, 13 pages. [cited by applicant]
Didyk, P. et al., “Adaptive Image-space Stereo View Synthesis,” Proceedings of VMV 2010 : Vision, Modeling, and Visualization, Nov. 15, 2010, Siegen, Germany, 8 pages. [cited by applicant]
Scherzer, D. et al., “A Survey on Temporal Coherence Methods in Real-Time Rendering,” Proceedings of the 32nd Annual Conference of the European Association for Computer Graphics, Eurographics 2011—State of the Art Repor… [cited by applicant]
Bowles, H. et al., “Iterative Image Warping,” Computer Graphics Fourm, vol. 31, No. 2, May 2012, 10 pages. [cited by applicant]
He, K. et al., “Fast Guided Filter,” ArXiv Cornell University Website, Available Online at https://arxiv.org/abs/1505.00996, May 5, 2015, 2 pages. [cited by applicant]
Schollmeyer, A. et al., “Efficient Hybrid Image Warping for High Frame-Rate Stereoscopic Rendering,” IEEE Transactions on Visualization and Computer Graphics, vol. 23, No. 4, Apr. 2017, Available Online Jan. 25, 2017, 1… [cited by applicant]
Yeh, R. et al., “Semantic Image Inpainting With Deep Generative Models,” Proceedings of the 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Jul. 21, 2017, Honolulu, Hawaii, 9 pages. [cited by applicant]
He, J. et al., “Hierarchical prediction-based motion vector refinement for video frame-rate up-conversion,” Journal of Real-Time Image Processing, vol. 17, Mar. 30, 2018, 15 pages. [cited by applicant]