IP Library › Granted Patent US 12,646,141
Granted Patent B2
US 12,646,141 · App. 18/506,755 · Granted Jun 2, 2026

Image processing apparatus and image processing method thereof

Inventors: Cheon Lee (Suwon-si, KR); Donghyun Kim (Suwon-si, KR); Yongsup Park (Suwon-si, KR); Jaeyeon Park (Suwon-si, KR); Iljun Ahn (Suwon-si, KR); Hyunseung Lee (Suwon-si, KR); Taegyoung Ahn (Suwon-si, KR); Youngsu Moon (Suwon-si, KR); Tammy Lee (Suwon-si, KR)
Assignee: SAMSUNG ELECTRONICS CO., LTD.
G06T3/403G06T3/4053G06T7/13G06V10/82G06T2207/20081
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 12,646,141
App. No.
18/506,755
Granted
Jun 2, 2026
Kind
B2
Abstract

An image processing apparatus applies an image to a first learning network model to optimize the edges of the image, applies the image to a second learning network model to optimize the texture of the image, and applies a first weight to the first image and a second weight to the second image based on information on the edge areas and the texture areas of the image to acquire an output image.

Claims (71)

1 . An image processing apparatus comprising:

a memory storing at least one instruction; and

a processor configured to execute the at least one instruction stored in the memory to:

obtain a first image in which a first area is processed and a first image characteristic is enhanced in an input image by applying the input image to a first neural network model,

obtain a second image in which a second area is processed and a second image characteristic is enhanced in the input image by applying the input image to a second neural network model,

obtain a first weight corresponding to the first image and a second weight corresponding to the second image by applying the input image to a third neural network model, and

provide an output image by mixing the first and second images according to the first weight and the second weight,

wherein each of the first neural network model and the second neural network model is trained to enhance different characteristics of an image.

2 . The image processing apparatus of claim 1 , wherein the processor is further configured to identify a plurality of objects included in the input image,

wherein the first area corresponds to a first object among the plurality of objects, and

wherein the second area corresponds to a second object among the plurality of objects.

3 . The image processing apparatus of claim 1 , wherein the processor is further configured to:

identify the first area included in the input image,

obtain the first image in which the first image characteristic of the first area is enhanced in the input image by using the first neural network model,

identify the second area included in the input image,

obtain the second image in which the second image characteristic of the second area is enhanced in the input image by using the second neural network model, and

provide the output image by mixing the first image and the second image.

4 . The image processing apparatus of claim 3 , wherein the first image characteristic of the first area corresponds to a characteristic of a graphic image, and

wherein the second image characteristic of the second area corresponds to a characteristic of a natural image.

5 . The image processing apparatus of claim 1 ,

wherein the first weight corresponds to the first area by,

wherein the second weight corresponds to the second area, and

wherein the processor is further configured to:

apply the first weight to the first image,

apply the second weight to the second image, and

mix the first image to which the first weight is applied and the second image to which the second weight is applied to provide the output image.

6 . The image processing apparatus of claim 1 , wherein the processor is further configured to:

classify each of a plurality of image blocks constituting the input image into either a graphic image or a natural image,

obtain the first image in which the first image characteristic of an image block classified as the graphic image from among the plurality of image blocks is enhanced as the first image by using the first neural network model, and

obtain the second image in which the second image characteristic of an image block classified as the natural image from among the plurality of image blocks is enhanced as the second image by using the second neural network model.

7 . The image processing apparatus of claim 6 , wherein the graphic image includes any one of an illustration image, a computer graphic image, or an animation image, and

wherein the natural image includes any one of a landscape image or a human image.

8 . The image processing apparatus of claim 6 , wherein the image block classified as the graphic image among the plurality of image blocks corresponds to the first area in the input image, and the image block classified as the natural image among the plurality of image blocks corresponds to the second area in the input image.

9 . The image processing apparatus of claim 1 , wherein the first neural network model and the second neural network model are different types of neural network models.

10 . The image processing apparatus of claim 1 , wherein the first neural network model is one of a deep learning model for enhancing the first image using a plurality of layers or a machine learning model trained to enhance the first image characteristic using a plurality of pre-learned filters, and

wherein the second neural network model is one of a deep learning model for enhancing the second image characteristic using a plurality of layers or a machine learning model trained to enhance the second image characteristic using a plurality of pre-learned filters.

11 . A method of controlling an image processing apparatus, the method comprising:

obtaining a first image in which a first area is processed and a first image characteristic is enhanced in an input image by applying the input image to a first neural network model;

obtaining a second image in which a second area is processed and a second image characteristic is enhanced in the input image by applying the input image to a second neural network model;

obtaining a first weight corresponding to the first image and a second weight corresponding to the second image by applying the input image to a third neural network model, and

providing an output image by mixing the first and second images according to the first weight and the second weight,

wherein each of the first neural network model and the second neural network model is trained to enhance different characteristics of an image.

12 . The method of claim 11 , wherein the method further comprises:

identifying a plurality of objects included in the input image,

wherein the first area corresponds to a first object among the plurality of objects, and

wherein the second area corresponds to a second object among the plurality of objects.

13 . The method of claim 11 , wherein the method further comprises:

identifying the first area included in the input image;

identifying the second area included in the input image,

wherein the obtaining the first image comprises obtaining the first image in which a first characteristic of the first area is enhanced in the input image by using the first neural network model,

wherein the obtaining the second image comprises obtaining the second image in which a second characteristic of the second area is enhanced in the input image by using the second neural network model, and

wherein the providing the output image comprises providing the output image by mixing the first image and the second image.

14 . The method of claim 13 , wherein the first characteristic of the first area corresponds to a characteristic of a graphic image, and

wherein the second characteristic of the second area corresponds to a characteristic of a natural image.

15 . The method of claim 11 ,

wherein the first weight corresponds to the first area,

wherein the second weight corresponds to the second area,

wherein the method further comprises:

applying the first weight to the first image;

applying the second weight to the second image; and

mixing the first image to which the first weight is applied and the second image to which the second weight is applied to provide the output image.

16 . The method of claim 11 , wherein the method further comprises:

classifying each of a plurality of image blocks constituting the input image into either a graphic image or a natural image,

wherein the obtaining comprises obtaining the first image in which the first image characteristic of an image block classified as the graphic image from among the plurality of image blocks is enhanced as the first image by using the first neural network model, and

wherein the obtaining comprises obtaining the second image in which the second image characteristic of an image block classified as the natural image from among the plurality of image blocks is enhanced as the second image by using the second neural network model.

17 . The method of claim 16 , wherein the graphic image includes any one of an illustration image, a computer graphic image, or an animation image, and

wherein the natural image includes any one of a landscape image or a human image.

18 . The method of claim 16 , wherein the image block classified as the graphic image among the plurality of image blocks corresponds to the first area in the input image, and the image block classified as the natural image among the plurality of image blocks corresponds to the second area in the input image.

19 . The method of claim 11 , wherein the first neural network model and the second neural network model are different types of neural network models.

20 . The method of claim 11 , wherein the first neural network model is one of a deep learning model for enhancing the first image characteristic using a plurality of layers or a machine learning model trained to enhance the first image characteristic using a plurality of pre-learned filters, and

wherein the second neural network model is one of a deep learning model for enhancing the second image characteristic using a plurality of layers or a machine learning model trained to enhance the second image characteristic using a plurality of pre-learned filters.

Priority Claims (2)
KR 10-2019-0060240 · May 22, 2019 · national
KR 10-2019-0080346 · Jul 3, 2019 · national
Continuity (3)
Continuation 17527732 · Nov 16, 2021
Continuation 16838650 · Apr 2, 2020
Related Publication 20240078631A1 · Mar 7, 2024
References Cited (58)
US 6233365B1 · Teruhiko · 2001 [cited by examiner]
US 8498450B2 · Nakanishi · 2013 [cited by applicant]
US 9280804B2 · Krishnaswamy et al. · 2016 [cited by applicant]
US 10007972B2 · Choi et al. · 2018 [cited by applicant]
US 10165204B2 · Zhou et al. · 2018 [cited by applicant]
US 10181092B2 · Shao et al. · 2019 [cited by applicant]
US 10489887B2 · El-Khamy et al. · 2019 [cited by applicant]
US 10650283B2 · Lee et al. · 2020 [cited by applicant]
US 10929749B2 · Srinivasa et al. · 2021 [cited by applicant]
US 10970820B2 · El-Khamy et al. · 2021 [cited by applicant]
US 11403838B2 · Zhu · 2022 [cited by examiner]
US 20030143634A1 · Terry · 2003 [cited by examiner]
US 20040184657A1 · Lin · 2004 [cited by examiner]
US 20180068463A1 · Risser · 2018 [cited by examiner]
US 20180075581A1 · Shi · 2018 [cited by examiner]
US 20180158240A1 · Saito · 2018 [cited by examiner]
US 20180293707A1 · El-Khamy et al. · 2018 [cited by applicant]
US 20180307983A1 · Srinivasa et al. · 2018 [cited by applicant]
US 20190188539A1 · Lee et al. · 2019 [cited by applicant]
US 20190378242A1 · Zhang · 2019 [cited by examiner]
US 20200090305A1 · El-Khamy et al. · 2020 [cited by applicant]
US 20200211413A1 · Zhang · 2020 [cited by examiner]
US 20200320769A1 · Chen · 2020 [cited by examiner]
US 20210224953A1 · El-Khamy et al. · 2021 [cited by applicant]
US 20210279571A1 · Srinivasa et al. · 2021 [cited by applicant]
US 20210303243A1 · Li · 2021 [cited by examiner]
US 20230066219A1 · Yu · 2023 [cited by examiner]
CN 102750695B · 2015 [cited by applicant]
CN 108596841A · 2018 [cited by applicant]
CN 108734645A · 2018 [cited by applicant]
CN 109345485A · 2019 [cited by applicant]
CN 109934247A · 2019 [cited by applicant]
EP 3499416A1 · 2019 [cited by applicant]
JP 2009296328A · 2009 [cited by applicant]
KR 101040461B1 · 2011 [cited by applicant]
KR 101795271B1 · 2017 [cited by applicant]
KR 101882704B1 · 2018 [cited by applicant]
KR 1020180097342A · 2018 [cited by applicant]
KR 1020180114488A · 2018 [cited by applicant]
KR 1020190006741A · 2019 [cited by applicant]
TW 201837854A · 2018 [cited by applicant]
Communication issued on Nov. 25, 2024 by the China National Intellectual Property Administration in Chinese Patent Application No. 202010080412.6. [cited by applicant]
Chao Dong et al., “Accelerating the Super-Resolution Convolutional Neural Network”, arXiv:1608.00367v1, Aug. 1, 2016, pp. 1-17, 19 pages total. [cited by applicant]
Communication dated Aug. 27, 2021, from the Intellectual Property Office of Taiwan in Application No. 109112722. [cited by applicant]
Communication dated Aug. 30, 2021, issued by the Korean Intellectual Property Office in Korean Application No. 10-2019-0080346. [cited by applicant]
Communication dated Sep. 8, 2020 issued by the European Intellectual Property Office in counterpart European Application No. 20165972.9. [cited by applicant]
Communication issued Apr. 20, 2022 by the European Patent Office in counterpart European Patent Application No. 20165972.9. [cited by applicant]
Communication issued Mar. 21, 2023 by the Korean Intellectual Property Office in counterpart Korean Patent Application No. 10-2022-0072692. [cited by applicant]
International Search Report (PCT/ISA/210) dated Aug. 14, 2020 issued by the International Searching Authority in counterpart International Application No. PCT/KR2020/006355. [cited by applicant]
International Written Opinion (PCT/ISA/237) dated Aug. 14, 2020 issued by the International Searching Authority in counterpart International Application No. PCT/KR2020/006355. [cited by applicant]
Kenji Suzuki et al., “Neural Edge Enhancer for Supervised Edge Enhancement from Noisy Images”, IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 25, No. 12, IEEE Computer Society, ISSN: 0162-8828, DOI… [cited by applicant]
Peng Ren et al., “Clustering-oriented Multiple Convolutional Neural Networks for Single Image Super-resolution”, ORE, Oct. 24, 2017, pp. 1-13 (14 pages total). [cited by applicant]
Sumei Li et al., “A two-channel convolutional neural network for image super-resolution”, Neurocomputing, vol. 275, No. 31, Jan. 31, 2018, pp. 267-277, 13 pages total. [cited by applicant]
Yang Zhao et al., “High Resolution Local Structure-Constrained Image Upsampling”, IEEE Transactions on Image Processing, vol. 24, No. 11, ISSN: 1057-7149, DOI: 10.1109/TIP.2015.2456416, Nov. 2015, pp. 4394-4407, 14 page… [cited by applicant]
Kwanyoung Kim et al., “SREdgeNET: Edge Enhanced Single Image Super Resolution using Dense Edge Detection Network and Feature Merge Network”, arxiv.org, Cornell University Library, Dec. 18, 2018, 10 pages total, XP080994… [cited by applicant]
Peng Ren eta al., “Clustering-Oriented Multiple Convolutional Neural Networks for Single Image Super-Resolution”, Cognitive Computation, vol. 10, No. 1, ISSN: 1866-9956, DOI: 10. 1007/S12559-017-9512-2, Oct. 4, 2017, pp… [cited by applicant]
Communication issued on Jun. 28, 2024 by the China National Intellectual Property Administration for Chinese Patent Application No. 202010080412.6. [cited by applicant]
Communication dated Feb. 10, 2025, issued by the Chinese National Intellectual Property Administration in Chinese Patent Application No. 202010080412.6. [cited by applicant]