WO2022191474A1 - 이미지의 화질을 개선하는 전자 장치 및 이를 이용한 이미지의 화질 개선 방법 - Google Patents
이미지의 화질을 개선하는 전자 장치 및 이를 이용한 이미지의 화질 개선 방법 Download PDFInfo
- Publication number
- WO2022191474A1 WO2022191474A1 PCT/KR2022/002649 KR2022002649W WO2022191474A1 WO 2022191474 A1 WO2022191474 A1 WO 2022191474A1 KR 2022002649 W KR2022002649 W KR 2022002649W WO 2022191474 A1 WO2022191474 A1 WO 2022191474A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- image
- person
- artificial intelligence
- intelligence model
- electronic device
- Prior art date
Links
- 238000000034 method Methods 0.000 title claims abstract description 45
- 238000013473 artificial intelligence Methods 0.000 claims abstract description 357
- 230000006872 improvement Effects 0.000 claims abstract description 128
- 238000012545 processing Methods 0.000 claims abstract description 41
- 230000001815 facial effect Effects 0.000 claims description 86
- 230000013016 learning Effects 0.000 claims description 66
- 230000015556 catabolic process Effects 0.000 claims description 3
- 238000006731 degradation reaction Methods 0.000 claims description 3
- 239000013585 weight reducing agent Substances 0.000 claims description 3
- 230000006870 function Effects 0.000 description 26
- 238000010586 diagram Methods 0.000 description 20
- 238000012549 training Methods 0.000 description 16
- 238000005516 engineering process Methods 0.000 description 13
- 238000001514 detection method Methods 0.000 description 11
- 238000013528 artificial neural network Methods 0.000 description 9
- 238000004891 communication Methods 0.000 description 6
- 230000006866 deterioration Effects 0.000 description 6
- 238000007781 pre-processing Methods 0.000 description 6
- 230000008569 process Effects 0.000 description 6
- 210000005069 ears Anatomy 0.000 description 4
- 210000004709 eyebrow Anatomy 0.000 description 4
- 210000004209 hair Anatomy 0.000 description 4
- 230000033001 locomotion Effects 0.000 description 4
- 238000010801 machine learning Methods 0.000 description 4
- 239000013598 vector Substances 0.000 description 4
- 230000037303 wrinkles Effects 0.000 description 4
- 238000013135 deep learning Methods 0.000 description 3
- 230000009471 action Effects 0.000 description 2
- 230000006835 compression Effects 0.000 description 2
- 238000007906 compression Methods 0.000 description 2
- 238000013527 convolutional neural network Methods 0.000 description 2
- 230000000306 recurrent effect Effects 0.000 description 2
- 230000004044 response Effects 0.000 description 2
- 230000000007 visual effect Effects 0.000 description 2
- 230000035045 associative learning Effects 0.000 description 1
- 230000002457 bidirectional effect Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000012790 confirmation Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 230000002452 interceptive effect Effects 0.000 description 1
- 230000004807 localization Effects 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 238000000691 measurement method Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000003058 natural language processing Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000013138 pruning Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 230000002787 reinforcement Effects 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 238000003786 synthesis reaction Methods 0.000 description 1
- 238000013519 translation Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/172—Classification, e.g. identification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/70—Denoising; Smoothing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformations in the plane of the image
- G06T3/40—Scaling of whole images or parts thereof, e.g. expanding or contracting
- G06T3/4053—Scaling of whole images or parts thereof, e.g. expanding or contracting based on super-resolution, i.e. the output image resolution being higher than the sensor resolution
- G06T3/4076—Scaling of whole images or parts thereof, e.g. expanding or contracting based on super-resolution, i.e. the output image resolution being higher than the sensor resolution using the original low-resolution images to iteratively correct the high-resolution images
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformations in the plane of the image
- G06T3/40—Scaling of whole images or parts thereof, e.g. expanding or contracting
- G06T3/4046—Scaling of whole images or parts thereof, e.g. expanding or contracting using neural networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/60—Image enhancement or restoration using machine learning, e.g. neural networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/168—Feature extraction; Face representation
- G06V40/171—Local features and components; Facial parts ; Occluding parts, e.g. glasses; Geometrical relationships
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N1/00—Scanning, transmission or reproduction of documents or the like, e.g. facsimile transmission; Details thereof
- H04N1/46—Colour picture communication systems
- H04N1/56—Processing of colour picture signals
- H04N1/60—Colour correction or control
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/102—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or selection affected or controlled by the adaptive coding
- H04N19/117—Filters, e.g. for pre-processing or post-processing
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/85—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using pre-processing or post-processing specially adapted for video compression
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10024—Color image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20092—Interactive image processing based on input by user
- G06T2207/20104—Interactive definition of region of interest [ROI]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30168—Image quality inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30196—Human being; Person
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30196—Human being; Person
- G06T2207/30201—Face
Definitions
- the disclosed embodiments relate to an electronic device for improving image quality and a method for improving image quality using the same.
- An artificial intelligence (AI) system is a computer system that implements human-level intelligence, and unlike the existing rule-based smart system, the machine learns, judges, and becomes smarter by itself. As artificial intelligence systems are used, the recognition rate improves and users can understand user preferences more accurately, and the existing rule-based smart systems are gradually being replaced by deep learning-based artificial intelligence systems.
- Machine learning Deep learning
- elemental technologies using machine learning.
- Machine learning is an algorithm technology that classifies/learns characteristics of input data by itself
- element technology is a technology that uses machine learning algorithms such as deep learning, such as language understanding, visual understanding, reasoning/prediction, knowledge expression, motion control, etc. It consists of technical fields of
- Linguistic understanding is a technology that recognizes and applies/processes human language/character. Natural Language Processing, Machine Translation, Dialog System, Question Answering, and Speech Recognition /Speech Recognition/Synthesis, etc.
- Visual understanding is a technology that recognizes and processes objects as if they were human eyes. Object Recognition, Object Tracking, Image Retrieval, Human Recognition, Scene Recognition , spatial understanding (3D Reconstruction/Localization), image enhancement, and the like.
- Inference prediction is a technology for logically reasoning and predicting information by judging information, such as Knowledge-based Reasoning, Optimization Prediction, Preference-based Planning, Recommendation, etc. includes
- Knowledge expression is a technology that automatically processes human experience information into knowledge data, and includes knowledge construction (data generation/classification) and knowledge management (data utilization).
- Motion control is a technology for controlling autonomous driving of a vehicle and movement of a robot, and includes motion control (navigation, collision, driving), manipulation control (action control), and the like.
- artificial intelligence technology can be used to acquire images such as photos or videos and to improve image quality.
- the method for generating a second high-quality person image by an electronic device disclosed as a technical means for achieving the above-described technical problem performing image processing on a low-quality first person image using an artificial intelligence model, Recognizing a person image, applying the first person image to an artificial intelligence model, and acquiring the second person image output from the artificial intelligence model, wherein the artificial intelligence model is a face recognition artificial intelligence model.
- the first face is recognized by performing face identification and face recognition on the first person image, and the image quality is improved for the area corresponding to the first face using the image quality improvement artificial intelligence model.
- a second person image may be obtained, and the second person image may be output to the electronic device.
- FIG. 1 is a diagram illustrating that an electronic device acquires a high-quality person image from a low-quality person image, according to an exemplary embodiment.
- FIG. 2 is a flowchart of a method for obtaining, by an electronic device, a high-quality person image from a low-quality person image, according to an exemplary embodiment.
- FIG. 3 is a diagram illustrating that an electronic device applies human images to an artificial intelligence model as training data, according to an exemplary embodiment.
- FIG. 4 is a diagram illustrating that an electronic device applies human images to an artificial intelligence model as training data, according to an exemplary embodiment.
- FIG. 5 is a diagram illustrating that an electronic device acquires facial features from a person image using an artificial intelligence model, according to an embodiment.
- FIG. 6 is a diagram illustrating that an electronic device performs image quality improvement on an image quality improvement area received from a user, according to an embodiment.
- FIG. 7 is a flowchart of a method of improving, by an electronic device, an image quality with respect to an image quality improvement area received from a user, according to an exemplary embodiment.
- FIG. 8 is a diagram illustrating that an electronic device improves image quality according to an image quality improvement direction received from a user, according to an embodiment.
- FIG. 9 is a flowchart of a method for improving picture quality according to a picture quality improvement direction received from a user by an electronic device, according to an exemplary embodiment.
- FIG. 10 is a block diagram of an electronic device, according to an embodiment.
- FIG. 11 is a block diagram of a software module stored in a memory of an electronic device, according to an exemplary embodiment.
- FIG. 12 is a block diagram of a server, according to an embodiment.
- FIG. 13 is a block diagram of a software module stored in a memory of a server, according to an embodiment.
- the method for generating a second high-quality person image by an electronic device disclosed as a technical means for achieving the above-described technical problem performing image processing on a low-quality first person image using an artificial intelligence model, Recognizing a person image, applying the first person image to an artificial intelligence model, and acquiring the second person image output from the artificial intelligence model, wherein the artificial intelligence model is a face recognition artificial intelligence model.
- the first face is recognized by performing face identification and face recognition on the first person image, and the image quality is improved for the area corresponding to the first face using the image quality improvement artificial intelligence model.
- a second person image may be obtained, and the second person image may be output to the electronic device.
- An electronic device for generating a high-quality second person image by performing image processing on a low-quality first person image by using an artificial intelligence model disclosed as a technical means for achieving the above-described technical problem, at least one command a memory for storing and a processor executing the at least one instruction, wherein the processor executes the at least one instruction to identify the first person image, apply the first person image to an artificial intelligence model, and , obtains the second person image output from the artificial intelligence model, and the artificial intelligence model uses the facial recognition artificial intelligence model to perform face identification and face recognition on the first person image, thereby performing a first By recognizing a face and performing image processing to improve image quality on an area corresponding to the first face using an image quality improvement artificial intelligence model, a second person image is obtained, and the second person image is converted into the electronic It can be output to the device.
- a computer-readable recording medium may record a program for executing at least one of the embodiments of the disclosed method in a computer.
- the application stored in the recording medium may be for executing at least one function among the disclosed method embodiments.
- 'part' as used herein may be a hardware component such as a processor or circuit, and/or a software component executed by a hardware component such as a processor, According to examples, a plurality of 'units' may be implemented as one unit (element), or one 'unit' may include a plurality of elements.
- Some embodiments of the present disclosure may be represented by functional block configurations and various processing steps. Some or all of these functional blocks may be implemented in various numbers of hardware and/or software configurations that perform specific functions.
- the functional blocks of the present disclosure may be implemented by one or more microprocessors, or by circuit configurations for a given function.
- the functional blocks of the present disclosure may be implemented in various programming or scripting languages.
- the functional blocks may be implemented as an algorithm running on one or more processors.
- the present disclosure may employ prior art for electronic configuration, signal processing, and/or data processing, and the like. Terms such as “mechanism”, “element”, “means” and “configuration” may be used broadly and are not limited to mechanical and physical configurations.
- connecting lines or connecting members between the components shown in the drawings only exemplify functional connections and/or physical or circuit connections.
- a connection between components may be represented by various functional connections, physical connections, or circuit connections that are replaceable or added.
- first and second may be used to describe various components, but the components should not be limited by the terms.
- the above terms may be used for the purpose of distinguishing one component from another.
- first data and the second data are described in this specification, they are only used to distinguish different data, and thus should not be limited thereto.
- the electronic device may use an artificial intelligence model to generate a high-quality image from a low-quality image.
- Functions related to artificial intelligence according to the present disclosure are operated through a processor and a memory.
- the processor may consist of one or a plurality of processors.
- the one or more processors may be a general-purpose processor such as a CPU, an AP, a digital signal processor (DSP), or the like, a graphics-only processor such as a GPU, a VPU (Vision Processing Unit), or an artificial intelligence-only processor such as an NPU.
- One or a plurality of processors control to process input data according to a predefined operation rule or artificial intelligence model stored in the memory.
- the AI-only processor may be designed with a hardware structure specialized for processing a specific AI model.
- the processor may perform a preprocessing process of converting data applied to the AI model into a form suitable for application to the AI model.
- AI models can be created through learning.
- being made through learning means that a basic artificial intelligence model is learned using a plurality of learning data by a learning algorithm, so that a predefined action rule or artificial intelligence model set to perform a desired characteristic (or purpose) is created means burden.
- Such learning may be performed in the device itself on which artificial intelligence according to the present disclosure is performed, or may be performed through a separate server and/or system.
- Examples of the learning algorithm include, but are not limited to, supervised learning, unsupervised learning, semi-supervised learning, or reinforcement learning.
- the artificial intelligence model may be composed of a plurality of neural network layers.
- Each of the plurality of neural network layers has a plurality of weight values, and a neural network operation is performed through an operation between an operation result of a previous layer and a plurality of weights.
- the plurality of weights of the plurality of neural network layers may be optimized by the learning result of the artificial intelligence model. For example, a plurality of weights may be updated so that a loss value or a cost value obtained from the artificial intelligence model during the learning process is reduced or minimized.
- the artificial neural network may include a deep neural network (DNN), for example, a Generative Adversarial Network (GAN), a Convolutional Neural Network (CNN), a Deep Neural Network (DNN), a Recurrent Neural Network (RNN), Restricted Boltzmann Machine (RBM), Deep Belief Network (DBN), Bidirectional Recurrent Deep Neural Network (BRDNN), or Deep Q-Networks, but is not limited to the above-described example.
- DNN Deep neural network
- GAN Generative Adversarial Network
- CNN Convolutional Neural Network
- DNN Deep Neural Network
- RNN Restricted Boltzmann Machine
- DBN Deep Belief Network
- BNN Bidirectional Recurrent Deep Neural Network
- Deep Q-Networks Deep Q-Networks
- the disclosed artificial intelligence model may be generated by learning a plurality of text data and image data input as learning data according to a predetermined criterion.
- the artificial intelligence model may generate result data by performing a learned function in response to input data, and may output the result data.
- the disclosed artificial intelligence model may include a plurality of artificial intelligence models trained to perform at least one function.
- FIG. 1 is a diagram illustrating that an electronic device acquires a high-quality person image from a low-quality person image, according to an exemplary embodiment.
- the electronic device 10 may generate a high-quality second person image by performing image processing on a low-quality first person image using an artificial intelligence model.
- the electronic device 10 may store the generated second person image in the memory 17 or output it to the display unit 12 - 1 .
- the low-resolution image is usually difficult to identify, or is degraded, such as an image in which the number of pixels is smaller than a predetermined number, an image in which the boundary is blurred by noise contained in the image, and an image in which the color temperature and hue are incorrectly specified. means image.
- the high-quality image refers to an image that is generally easy to identify, such as an image including the number of pixels greater than or equal to a predetermined number, an image with clear boundaries included in the image, and an image with accurate color temperature and color tone.
- enhancement of image quality means improving the deterioration factor of an image.
- image quality enhancement includes image processing such as resolution enhancement, noise reduction, artifact removal, color adjustment, and sharpness enhancement.
- the electronic device 10 may identify a low-quality person image.
- the electronic device 10 may apply a low-quality human image to the AI model.
- the artificial intelligence model may be trained to generate and output a high-quality person image by performing image processing on a low-quality person image.
- the electronic device 10 may store the high-definition person image output from the artificial intelligence model in the memory 17 or may output it through the display unit 12-1.
- the artificial intelligence model may be built in at least one of the electronic device 10 and the server 20 .
- an artificial intelligence model built in the electronic device 10 will be described as an example, but the present invention is not limited thereto.
- the artificial intelligence model built in the electronic device 10 to be described below may be applied by analogy to the artificial intelligence model built in the server 20 .
- the electronic device 10 includes a mobile device (eg, a smart phone, a tablet PC, etc.) capable of transmitting/receiving data to and from the server 20 through a network, a general-purpose computer (PC, Personal Computer), and The same computing device may be included.
- the electronic device 10 includes an Internet of Things (ioT) device, various Internet of Things devices, and a home hub device (eg, a router, an interactive artificial intelligence speaker, etc.) connected to the server 20 . can do.
- the electronic device 10 includes a computing device such as a mobile device (eg, a smart phone, a tablet PC, etc.), a general computer (PC, Personal Computer), and a server on which the artificial intelligence model 19 is built. may include
- the electronic device 10 may perform predetermined operations using the artificial intelligence model 19 .
- the electronic device 10 may perform operations of identifying and classifying input data using the artificial intelligence model 19 , and outputting data corresponding to the input data.
- the server 20 may transmit/receive data to and from the electronic device 10 .
- the server 20 may apply data received from the electronic device 10 as the AI model 29 , and transmit data output from the AI model 29 to the electronic device 10 .
- the server 20 may transmit data used to update the artificial intelligence model 19 built in the electronic device 10 to the electronic device 10 .
- the artificial intelligence model may be composed of a plurality of artificial intelligence models trained to perform a predetermined function.
- the AI model is a preprocessing AI model that performs preprocessing as a type applicable to the AI model, a picture quality classification AI model that classifies the picture quality of a person image applied to the AI model, and at least one A face detection AI model that detects a face, a face recognition AI model that identifies a person corresponding to a face identified from a person image, and image processing on a low-quality person image It may include, but is not limited to, an image quality improvement artificial intelligence model for generating a person image of There may be a plurality of artificial intelligence models performing the same function, and one AI model may perform at least one or more functions of the disclosed embodiments.
- the artificial intelligence model may perform learning of the artificial intelligence model using learning data.
- the artificial intelligence model may use a plurality of person images as training data to learn the face identification AI model, the face recognition AI model, and the image quality improvement AI model.
- the artificial intelligence model may use a low-quality person image and a high-quality person image obtained by converting a high-quality person image as a pair of learning data.
- the artificial intelligence model may use a plurality of person images classified for each person as learning data.
- the artificial intelligence model may use training data for the image quality improvement part selected by the user.
- the artificial intelligence model may use training data for the image quality improvement direction selected by the user.
- the electronic device 10 may perform image quality improvement in response to an input of a user selecting an image.
- the electronic device 10 may perform image quality improvement by applying an image designated by the user to perform image quality improvement to the AI model.
- the electronic device 10 may improve the image quality of an image stored in the electronic device 10 during a time when the user is not using the electronic device 10 .
- the electronic device 10 may identify whether an image stored in the electronic device 10 is a low-quality image during idle time, and may perform image quality improvement.
- the electronic device 10 may acquire a high-quality image by identifying a low-quality image and performing image quality improvement.
- FIG. 2 is a flowchart of a method for obtaining, by an electronic device, a high-quality person image from a low-quality person image, according to an exemplary embodiment.
- the electronic device 10 may identify a low-quality first person image.
- the first person image stored in the memory 17 of the electronic device 10 may be identified.
- the electronic device 10 may identify the first person image obtained by using the camera of the electronic device 10 .
- the electronic device 10 may identify the first person image shared on the web.
- the electronic device 10 may identify the first person image shared through the application.
- the electronic device 10 may identify the first person image based on an input of a user selecting a low-quality person image. For example, the electronic device 10 may identify the first person image based on an input for selecting an image for which the user wants to improve image quality from among a plurality of person images.
- the electronic device 10 may identify a low-quality first person image based on a predetermined criterion. For example, the electronic device 10 may identify an image including the number of pixels less than a predetermined number and an image having a frequency with respect to a boundary line equal to or less than a predetermined value.
- the electronic device 10 may identify the first person image using the artificial intelligence model.
- the electronic device 10 may identify a low-quality first person image from a plurality of images by using a disclaimer artificial intelligence model.
- the discrimination AI model an AI model trained to distinguish a high-definition person image and a high-quality person image generated by the image quality improvement AI model from a low-quality person image may be used.
- the electronic device 10 may apply the first person image to the AI model.
- the electronic device 10 may apply the first person image identified in step S210 to the AI model.
- the electronic device 10 may apply the pre-processed first person image to the AI model. For example, the electronic device 10 may identify a face in the first person image and apply information about the result of identifying the face as an AI model together with the first person image. As another example, the electronic device 10 may recognize a face in the first person image and apply information about a person corresponding to the recognized face as an artificial intelligence model together with the first person image. have. As another example, the electronic device 10 may apply information about a result of classifying the first person image as a low-quality image together with the first person image as an artificial intelligence model based on a predetermined criterion. Alternatively, the artificial intelligence model may directly perform pre-processing on the input first person image without performing the pre-processing described above.
- the electronic device 10 may apply information about an input for selecting an image quality improvement part of a person included in the first person image received from the user to the AI model.
- the electronic device 10 may apply information about a user input that selects to increase the detail of the person's eyes to the AI model.
- the electronic device 10 may apply information about an input for selecting a picture quality improvement direction of the first person image received from the user to the AI model.
- the electronic device 10 may apply information about a user's input selecting to revise the first person image similarly to a predetermined image to the AI model.
- the electronic device 10 may apply information about at least one of color, sharpness, and resolution of the image selected by the user to the AI model.
- the electronic device 10 may apply information about a user's input selecting to revise a person included in the first person image similarly to a predetermined person to the AI model.
- the electronic device 10 may transmit the first person image to the server 20 in order to apply the first person image to the artificial intelligence model built in the server 20 .
- the artificial intelligence model may be learned from a plurality of person images as learning data so as to perform face identification and face recognition from the person image and improve image quality on the recognized face.
- the artificial intelligence model may be a pair of learning of a high-quality person image and a low-quality person image to which deterioration is applied to the high-quality person image so as to improve the quality of the low-quality person image.
- the AI model may be a personalized learning for each of the people by using a plurality of person images classified for each person as learning data.
- the artificial intelligence model on which the personalized learning has been performed may be weight-reduced.
- the artificial intelligence model may be learned to acquire facial features for each person by performing personalized learning.
- the AI model may be learned to identify and recognize a corresponding person from a person image based on the acquired facial features.
- the artificial intelligence model may be trained to improve image quality based on the acquired facial features.
- the artificial intelligence model may be trained to improve image quality by performing image processing on a region on the image corresponding to the image quality improvement region selected by the user based on the acquired facial features.
- the artificial intelligence model may be learned to improve image quality by modifying acquired facial features according to the image quality improvement direction selected by the user.
- the electronic device 10 may obtain a high-quality second person image output from the artificial intelligence model.
- the electronic device 10 may obtain the second person image obtained by the artificial intelligence model performing image quality improvement on the first person image from the artificial intelligence model.
- the electronic device 10 may acquire the second person image generated by performing image quality improvement on the face recognized from the first person image.
- the electronic device 10 may obtain the second person image generated by performing image processing for increasing the resolution of the first person image.
- the electronic device 10 may obtain the second person image generated by performing image processing for removing noise on the first person image.
- the electronic device 10 may acquire the generated second person image by performing image processing for adjusting the color of the first person image.
- the electronic device 10 may acquire the generated second person image by performing image processing for adjusting the color of the first person image.
- the electronic device 10 may acquire the generated second person image by performing image processing for improving the sharpness of the first person image.
- the electronic device 10 may obtain, from the artificial intelligence model, the second person image in which the image quality has been improved with respect to the region corresponding to the image quality improvement portion selected by the user.
- the electronic device 10 may apply information about a user input that selects to increase the detail of the person's eyes to the AI model.
- the electronic device 10 may obtain the second person image, on which the image quality is improved, from the AI model according to the image quality improvement direction selected by the user.
- the electronic device 10 may acquire the second person image on which image quality has been improved according to a user's input selecting to correct the first person image similarly to a predetermined image.
- the electronic device 10 may acquire a second person image whose image quality has been improved similarly to at least one of color, sharpness, and resolution of the image selected by the user.
- the electronic device 10 may acquire a second person image in which a person included in the first person image is modified to be similar to a predetermined person according to a user input.
- the electronic device 10 may receive the second person image generated from the artificial intelligence model built in the server 20 .
- the electronic device 10 may display the second person image on the display unit 12-1.
- the electronic device 10 may display the first person image and the second person image together so that they can be compared.
- the electronic device 10 may store the second person image in the memory 17 according to a user's confirmation input for the image quality improvement result.
- the electronic device 10 may store the second person image in a location where the first person image is stored.
- the electronic device 10 may store the second person image together with the first person image.
- the electronic device 10 may replace the first person image with the second person image and store it.
- FIG. 3 is a diagram illustrating that an electronic device applies human images to an artificial intelligence model as training data, according to an exemplary embodiment.
- the electronic device 10 may apply a plurality of high-quality images 310 and a plurality of low-resolution images 330 to the AI model as training data.
- the electronic device 10 may identify a plurality of images stored in the memory 17 of the electronic device 10 and apply them to the AI model.
- the electronic device 10 may download a plurality of images disclosed on the web and apply them to the AI model.
- the electronic device 10 may apply a plurality of images shared through an application to the AI model.
- the electronic device 10 applies the plurality of low-resolution images 330 and the plurality of high-quality images 310, each of which are deteriorated of the plurality of high-quality images 310, to the AI model as training data. can do.
- the electronic device 10 uses the plurality of low-resolution images 330 and the plurality of high-quality images 310 obtained by performing down-sampling on each of the plurality of high-quality images 310 as an artificial intelligence model. can be applied to
- the electronic device 10 obtains a plurality of low-resolution images 330 and a plurality of high-quality images 310 obtained by performing image processing in which noise is applied to each of the plurality of high-quality images 310 . ) can be applied to the AI model.
- the electronic device 10 obtains a plurality of low-resolution images 330 and a plurality of high-quality images obtained by performing image processing for applying a blur to each of the plurality of high-quality images 310 .
- 310 can be applied to the artificial intelligence model.
- the electronic device 10 obtains a plurality of low-resolution images 330 and a plurality of high-quality images 310 obtained by performing image processing for changing each color of the plurality of high-quality images 310 .
- the electronic device 10 obtains a plurality of low-resolution images 330 and a plurality of high-quality images 330 obtained by performing image processing for changing each color of the plurality of high-quality images 310 . can be applied to AI models.
- the electronic device 10 may group a high-quality image and a low-resolution image in pairs and apply them to the AI model.
- the electronic device 10 groups a first high-quality image 311 and a first low-resolution image 313 obtained by degradation conversion of the first high-quality image 311 from among the plurality of high-quality images 310 into a pair. and the first high-resolution image 311 and the first low-resolution image 313 may be applied to the AI model together.
- the electronic device 10 may apply only the plurality of high-quality images 310 to the AI model.
- the artificial intelligence model 19 may acquire a plurality of low-quality images 330 by performing deterioration application on each of the plurality of high-quality images 310 .
- the artificial intelligence model 19 may perform learning by using a plurality of high-quality images 310 and a plurality of low-resolution images 330 as learning data.
- the electronic device 10 may perform pre-processing on the plurality of high-quality images 310 and the plurality of low-resolution images 330 and apply them to the artificial intelligence model. For example, the electronic device 10 identifies a face with respect to a plurality of high-quality images 310 and a plurality of low-resolution images 330 , extracts regions corresponding to the identified face, and uses the artificial intelligence model as training data. can be applied to
- the artificial intelligence model 19 may learn the applied first high-quality image 311 and the first low-resolution image 331 .
- the artificial intelligence model 19 may train the image quality improvement artificial intelligence model to generate the first high-quality image 311 from the first low-resolution image 331 .
- the artificial intelligence model 19 may train a discriminative artificial intelligence model for identifying a difference by comparing the second high-quality image generated from the first low-resolution image 331 with the first high-quality image 311 .
- the artificial intelligence model may train the image quality improvement artificial intelligence model so that the loss function of Equation 1 is applied.
- the loss function of the image quality improvement AI model is the loss function of the image quality improvement AI model, is the objective function (loss function) of the general GAN method artificial intelligence model, is an objective function (loss function) for the image quality improvement AI model to generate a high-quality image as closely as possible from a low-resolution image.
- the artificial intelligence model 19 may train the discrimination artificial intelligence model to distinguish the first high-quality image 311 from the second high-quality image by using an objective function (Loss Function).
- the electronic device 10 may receive information about the artificial intelligence model learned from the server 20 .
- the server 20 may train the artificial intelligence model built in the server 20 , and transmit data about the artificial intelligence model 19 updated by learning to the electronic device 10 .
- the electronic device 10 may receive information on the updated weight among the weights of the artificial intelligence model, and update the artificial intelligence model 19 built in the electronic device 10 using the received information. have.
- FIG. 4 is a diagram illustrating that an electronic device applies human images to an artificial intelligence model as training data, according to an exemplary embodiment.
- the electronic device 10 may create a database including various person images for a specific person by personalizing each of the plurality of person images to be included in one of the plurality of persons. . That is, the electronic device 10 may personalize the plurality of person images, such as the first person image 410 , the second person image 430 , and the third person image 450 . The electronic device 10 applies the classified first person image 410 , the second person image 430 , and the third person image 450 to the AI model as learning data, thereby performing personalized learning of the AI model. can
- the electronic device 10 may personalize the plurality of person images based on an input selected by the user as the same person.
- the electronic device 10 may personalize a plurality of person images based on a path in which the images are stored. For example, the electronic device 10 may personalize images stored in the same folder of the memory 17 as images relating to the same person.
- the electronic device 10 may personalize the plurality of person images based on the user who provided the image. For example, the electronic device 10 may personalize images provided from the same user into images relating to the same person by using a messenger application.
- the electronic device 10 may personalize a plurality of person images by using an artificial intelligence model. For example, the electronic device 10 may perform face detection from a plurality of person images using the face detection AI model. The electronic device 10 may identify the first face by performing face recognition on the detected face using the face recognition artificial intelligence model. The electronic device 10 may identify a first person corresponding to the first face from among a plurality of people based on the face recognition result.
- the electronic device 10 may apply information about the personalized classification result of the plurality of person images together with the plurality of person images to the AI model. For example, the electronic device 10 may insert a tag related to a personalized classification result in the plurality of high-quality person images and the plurality of low-quality person images.
- the electronic device 10 may apply a plurality of high-quality person images and a plurality of low-resolution person images into which tags are inserted to the AI model.
- the plurality of low-quality person images may be generated by applying deterioration to the plurality of high-quality person images.
- the electronic device 10 may apply a plurality of person images for which face identification is not performed to the AI model.
- the artificial intelligence model 19 may use a plurality of person images as training data to train the face identification AI model and the facial recognition AI model.
- the artificial intelligence model 19 may update the image quality improvement AI model by performing personalized learning on the image quality improvement AI model using a plurality of inputted person images.
- the updated image quality improvement AI model may be a specialized model to improve the image quality of a low-quality image of a specific person.
- the artificial intelligence model uses the first person image 410, the second person image 430, and the third person image 450 to learn the image quality improvement AI model, so that the first person, the second person and a specialized image quality improvement AI model to improve the image quality of the low-quality image for the third person may be acquired.
- the artificial intelligence model 19 may perform weight reduction on at least one of a face detection AI model for performing personalized learning, a face recognition AI model, and an image quality improvement AI model.
- the artificial intelligence model 19 may apply a weight reduction technique such as filter pruning to the image quality improvement artificial intelligence model.
- the artificial intelligence model 19 may obtain an image quality improvement artificial intelligence model having excellent image quality improvement performance for a specific person who has been individually trained and light-weighted data.
- the artificial intelligence model 19 may obtain a facial recognition artificial intelligence model with excellent facial recognition performance and light-weight data for a specific person who has been individually trained.
- FIG. 5 is a diagram illustrating that an electronic device acquires facial features from a person image using an artificial intelligence model, according to an embodiment.
- the electronic device 10 may acquire facial features of a person included in the person image 510 by applying the person image 510 to the artificial intelligence model 19 .
- the artificial intelligence model 19 may perform face detection from the person image 510 .
- the artificial intelligence model 19 may detect at least one face from the person image 510 using the face detection artificial intelligence model.
- the artificial intelligence model 19 may acquire facial features from the detected face.
- the artificial intelligence model 19 uses a facial recognition artificial intelligence model to detect facial contours and facial landmarks (eg, eyes, nose, mouth, ears, etc.) shape and size detected from a person image. , proportion and location, facial features such as facial details (eg, eyebrows, wrinkles, hair, skin tone).
- the AI model 19 may acquire facial features of a specific person by performing personalized learning by the facial recognition AI model.
- the artificial intelligence model 19 may acquire the facial features of the first person by learning the person image of the first person.
- the AI model 19 may update the facial recognition AI model using the acquired facial features.
- the AI model 19 may update the weight of the facial recognition AI model using facial features for a specific person.
- the artificial intelligence model 19 may update the image quality improvement artificial intelligence model using the acquired facial features.
- the artificial intelligence model 19 may include facial contours, facial landmarks (eg, eyes, nose, mouth, ears, etc.) shape, size and location, and facial details (eyebrows, wrinkles, hair). It is possible to train an image quality improvement AI model to correct facial features such as facial contours, facial landmarks (eg, eyes, nose, mouth, ears, etc.) shape, size and location, and facial details (eyebrows, wrinkles, hair). It is possible to train an image quality improvement AI model to correct facial features such as
- FIG. 6 is a diagram illustrating that the electronic device performs image quality improvement on an image quality improvement area received from a user, according to an embodiment. It is a flowchart of a method for improving image quality for
- the electronic device 10 may receive an input for selecting an image quality improvement part for a person included in a first person image 610 from a user.
- the electronic device 10 may output the second person image 630 in which the quality of the first person image 610 is improved based on an input received from the user.
- the electronic device 10 may receive an input for selecting an image enhancement region from the user 1 and may identify a region of image enhancement selected by the user.
- the electronic device 10 may receive, from the user 1 , a user input for selecting a part requiring image quality improvement from among the facial parts of a person included in the first person image 610 .
- the electronic device 10 may receive a user input for selecting eyes of a person included in the first person image 610 from the user 1 .
- the electronic device 10 may receive a user's input for selecting a part requiring image quality improvement provided as a list (preset) from among the facial parts of the person included in the first person image 610 .
- the electronic device 10 is illustrated to receive a user input for selecting an image quality improvement area based on a touch input of the user 1 , but the present invention is not limited thereto.
- the electronic device 10 may identify an image quality improvement area based on a user input received through various interfaces capable of receiving a user input.
- the electronic device 10 may detect the face of the first person and facial features of the first person from the first person image 610 using the face detection AI model.
- the electronic device 10 may identify a region corresponding to the image quality improvement region selected by the user 1 by using the face detection AI model.
- the electronic device 10 detects facial features using the face detection artificial intelligence model, and performs face parsing based on the detected facial features, so that the region where the user input is received is It can be identified that corresponds to the eyes of the first person.
- the electronic device 10 may identify the first person corresponding to the face detected from the first person image 610 by using the face recognition artificial intelligence model.
- the electronic device 10 may apply the information on the image quality improvement region to the artificial intelligence model 19 .
- the electronic device 10 may apply the information on the image quality improvement region to the artificial intelligence model 19 together with the first person image 610 .
- the electronic device 10 may apply the first person image 610 marked with the area where the user input is received to the artificial intelligence model 19 .
- the electronic device 10 may apply the feature information on the area where the user input is received to the artificial intelligence model 19 together with the first face image 610 .
- the electronic device 10 may apply the information on the facial region corresponding to the region where the user input is received to the artificial intelligence model 19 together with the first facial image 610 .
- the electronic device 10 uses the facial recognition artificial intelligence model to collect information about the first person identified from the first person image 610 together with the first person image 610 as an artificial intelligence model ( 19) can be applied.
- the electronic device 10 may apply a plurality of high-quality images including the first person together with the first person image 610 to the artificial intelligence model 19 .
- the electronic device 10 may train an artificial intelligence model by using the training data for the image quality improvement region.
- the artificial intelligence model 19 may identify the image quality improvement region from the first person image 610 based on the input image quality improvement region information. For example, the artificial intelligence model 19 may identify an image quality improvement region from the first person image 610 based on information on a facial region (eg, eyes) selected by the user 1 .
- a facial region eg, eyes
- the artificial intelligence model 19 may acquire learning data on the image quality improvement region based on the input information on the image quality improvement region. For example, the artificial intelligence model 19 obtains a plurality of high-quality images for the user-selected facial region (eg, eyes) based on information on the user-selected facial region (eg, eyes). can do. The artificial intelligence model 19 may output data requesting to apply a plurality of high-definition images to the electronic device 10 .
- the artificial intelligence model 19 may train the face detection artificial intelligence model to perform face parsing by learning the learning data input from the electronic device 10 . have.
- the artificial intelligence model 19 is based on the face parsing result data output from the face detection artificial intelligence model and information on the image quality improvement area selected by the user, the purpose of the image quality improvement artificial intelligence model You can set a Loss Function.
- the artificial intelligence model 19 may set a weighted loss objective function (loss function) for the image quality improvement artificial intelligence model.
- the artificial intelligence model 19 may train the image quality improvement artificial intelligence model by using the learning data input from the electronic device 10 .
- the artificial intelligence model 19 may learn the image quality improvement AI model by using a loss function set for the image quality improvement AI model and learning data input from the electronic device 10 .
- the artificial intelligence model 19 may train the image quality improvement artificial intelligence model using a plurality of high-definition images for a facial region (eg, eyes) selected by the user.
- the artificial intelligence model 19 uses a plurality of high-quality images for the first person included in the first person image 610 to determine the facial region (eg, eyes) of the first person selected by the user. It is possible to train a picture quality improvement artificial intelligence model that improves picture quality.
- the electronic device 10 may acquire a second person image with improved image quality in the image quality improvement area.
- the artificial intelligence model 19 may generate the second person image 630 by performing image quality improvement on the first person image 610 using the image quality improvement AI model updated in step S750.
- the artificial intelligence model 19 performs image quality improvement on a predetermined facial region (eg, eyes) of the first person included in the first person image 610 , so that the second person image 630 is ) can be created.
- the artificial intelligence model 19 may output the generated second person image 630 to the electronic device.
- the electronic device 10 may display the second person image on the display unit 12-1.
- the electronic device 10 may display the first person image and the second person image together so that they can be compared.
- the electronic device 10 may store the second person image 630 output from the artificial intelligence model 19 in the memory 17 .
- the electronic device 10 may store the second person image in a path in which the first person image is stored.
- the electronic device 10 may store the second person image together with the first person image.
- the electronic device 10 may replace the first person image with the second person image and store it.
- FIG. 8 is a diagram illustrating that the electronic device improves image quality according to the image quality improvement direction received from the user, according to an embodiment. Therefore, it is a flowchart of a method for improving image quality.
- the electronic device 10 generates a third person image 820 based on an input of the user 1 selecting the third person image 820 in the direction of improving the quality of the first person image 810 . ) from which feature information can be obtained.
- the electronic device 10 may acquire the second person image 830 by improving the quality of the first person image 810 by using the characteristic information obtained from the third person image 820 .
- the electronic device 10 performs image processing to increase the resolution of the first person image 810 so as to correspond to the resolution of the third person image 820 by using the artificial intelligence model 19, 2 A person image 830 may be acquired.
- the electronic device 10 performs image processing for adjusting the color of the first person image 810 to correspond to the color of the third person image 820 using the artificial intelligence model 19 .
- a second person image 830 may be acquired.
- the electronic device 10 performs image processing for adjusting the sharpness of the first person image 810 to correspond to the sharpness of the third person image 820 using the artificial intelligence model 19 .
- a second person image 830 may be acquired.
- the electronic device 10 uses the artificial intelligence model 19 to match the facial features of the second person included in the third person image 820 to the second person included in the first person image 810 .
- a second person image 830 may be obtained.
- the electronic device 10 may receive an input related to image quality improvement “Num ⁇ *” from the user.
- the electronic device 10 receives from the user an input regarding the image quality improvement direction for adjusting at least one of the resolution, sharpness, color, noise removal, and artifact removal generated during image compression of the first person image 810 . can receive
- the electronic device 10 provides information on the image quality improvement direction from the user through an interface that selects at least one of the resolution, sharpness, color, noise removal, and artifact removal generated during image compression of the first person image 810 . input can be received.
- the electronic device 10 may receive a user input for selecting the third person image 820 such that the first person image 810 includes image properties similar to those of the third person image 820 . have. That is, the electronic device 10 may receive a user input for selecting an image quality improvement direction so that the first person image 810 corresponds to at least one of resolution, sharpness, color, noise, and artifacts of the third person image 820 . can
- the electronic device 10 may receive an input from the user regarding the direction of image quality improvement for correcting the facial features of the first person included in the first person image 810 . Specifically, the electronic device 10 adjusts the facial features of the first person included in the first person image 810 to match the facial features of the second person included in the third person image 820 in the image quality improvement direction. input can be received.
- the electronic device 10 may apply the information on the image quality improvement direction to the AI model.
- the electronic device 10 may apply the information on the image quality improvement direction to the artificial intelligence model 19 together with the first person image 810 .
- the electronic device 10 provides information about at least one of the resolution, sharpness, color, noise removal, and artifact removal of the compressed image of the first person image 810 selected by the user to the first person image 810 . It can be applied to the artificial intelligence model (19) together with As another example, the electronic device 10 may apply the third person image 820 selected by the user to the artificial intelligence model 19 together with the first person image 810 .
- the electronic device 10 uses a plurality of high-definition data related to the second person included in the third person image 820 together with the first person image 810 as learning data as the artificial intelligence model 19 .
- the electronic device 10 may detect a face from the third person image 820 and identify a second person corresponding to the detected face.
- the electronic device 10 may acquire a plurality of high-quality images including the second person.
- the electronic device 10 generates a plurality of low-resolution images by performing deterioration application on each of the plurality of high-quality images including the second person, and stores the generated low-resolution images together with the plurality of high-quality images in the artificial intelligence model. It can be applied as learning data.
- the electronic device 10 may train the artificial intelligence model by using the learning data on the image quality improvement direction.
- the artificial intelligence model 19 may obtain feature vectors of the third person image 820 from the third person image 820 .
- the artificial intelligence model 19 may train the image quality improvement artificial intelligence model by using the feature vectors of the third person image 820 .
- the artificial intelligence model 19 acquires feature vectors related to at least one of the resolution, sharpness, color, noise, and artifacts of the third person image 820 , and uses the acquired feature vectors to select the image quality of the user.
- the image quality improvement AI model may be trained to improve the image quality of the first person image 810 .
- the artificial intelligence model 19 may acquire the facial features of the second person from the third person image 820 .
- the artificial intelligence model 19 uses the facial recognition artificial intelligence model to obtain the facial contour of the second person from the third person image 820, and facial landmarks (eg, eyes, nose, mouth, Facial features such as the shape, size and location of the ears) and facial details (eyebrows, wrinkles, hair, skin tone) can be acquired.
- facial landmarks eg, eyes, nose, mouth, Facial features such as the shape, size and location of the ears
- facial details eyebrows, wrinkles, hair, skin tone
- the artificial intelligence model 19 may acquire the facial features of the second person from a plurality of images applied together with the third person image 820 .
- the artificial intelligence model 19 may detect the face of the second person from each of a plurality of high-definition images including the second person, and obtain facial features of the second person.
- the artificial intelligence model 19 detects the face of the second person from each of a plurality of low-quality images generated by applying deterioration to a plurality of high-quality images including the second person, and facial features of the second person can be obtained.
- the artificial intelligence model 19 may train the image quality improvement artificial intelligence model by using the acquired facial features of the second person.
- the artificial intelligence model 19 includes the facial features of the second person image 830 and the third person image ( 820), it is possible to set an objective function (loss function) of the image quality improvement artificial intelligence model so that the facial features obtained from it are similar.
- the artificial intelligence model 19 may learn the image quality improvement artificial intelligence model using a loss function set for the image quality improvement AI model and learning data input from the electronic device 10 .
- the electronic device 10 may acquire the second person image according to the image quality improvement direction.
- the artificial intelligence model 19 uses the image quality improvement artificial intelligence model to correspond to the image quality improvement direction selected by the user to improve the resolution, sharpness, color adjustment, noise removal and
- the second person image 830 may be obtained by performing image processing of at least one of removing artifacts generated in the compressed image.
- the artificial intelligence model 19 may be applied to the first person image 810 so that the first person image 810 corresponds to at least one of the resolution, sharpness, color, noise, and artifacts of the third person image 820 . image processing can be performed.
- the artificial intelligence model 19 uses the image quality improvement artificial intelligence model to match the facial features of the second person included in the third person image 820 to the second person included in the first person image 810 .
- a second person image 830 may be obtained.
- the artificial intelligence model 19 may determine the facial landmarks of the first person according to the shape, size and proportion of the facial landmarks (eg, eyes, nose, mouth, ears, etc.) of the second person. You can adjust the shape, size and proportions.
- the artificial intelligence model 19 may adjust the facial details of the first person to match the facial details of the second person (eg, eyebrows, wrinkles, hair, skin tone, etc.).
- FIG. 10 is a block diagram of an electronic device, according to an embodiment.
- the electronic device 10 may include a user input unit 11 , an output unit 12 , a processor 13 , a communication unit 15 , and a memory 17 .
- the electronic device 10 may be implemented by more components than those illustrated in FIG. 10 , or the electronic device 10 may be implemented by fewer components than those illustrated in FIG. 10 .
- the user input unit 11 means a means for a user to input data for controlling the electronic device 10 .
- the user input unit 11 includes a touch screen, a key pad, a dome switch, a touch pad (contact capacitive method, pressure resistance film method, infrared sensing method, Surface ultrasonic conduction method, integral tension measurement method, piezo effect method, etc.), a touch screen, a jog wheel, a jog switch, etc. may be used, but are not limited thereto.
- the user input unit 11 may receive a user input necessary for the electronic device 10 to perform the embodiments described with reference to FIGS. 1 to 9 .
- the output unit 12 outputs information processed by the electronic device 10 .
- the output unit 12 may output information related to the embodiments described with reference to FIGS. 1 to 9 .
- the output unit 12 may include an object, a user interface, and a display unit 12-1 that displays a result of performing an operation corresponding to a user's input.
- the processor 13 generally controls the overall operation of the electronic device 10 .
- the processor 13 executes at least one instruction stored in the memory 17 , so that the user input unit 11 , the output unit 12 , the communication unit 15 , and the memory 17 perform associative learning. ) can be controlled in general.
- the processor 13 may control the electronic device 10 to detect a face from an image and identify a person corresponding to the detected face by executing instructions stored in the face identification and face recognition module 17a.
- the processor 13 may control the electronic device 10 to detect a face from an image and identify a person corresponding to the detected face by executing instructions stored in the face identification and face recognition module 17a.
- the processor 13 may control the electronic device 10 to acquire the facial feature detected from the image by executing an instruction stored in the facial feature acquiring module 17b.
- the content overlapping with the embodiment described above with reference to FIGS. 1 to 9 will be omitted.
- the processor 13 may control the electronic device 10 to improve the image quality by executing an instruction stored in the image quality improvement module 17c.
- the content overlapping with the embodiment described above with reference to FIGS. 1 to 9 will be omitted.
- the processor 13 may control the electronic device 10 so that the AI model learns the training data by executing the instructions stored in the AI model learning module 17d.
- the content overlapping with the embodiment described above with reference to FIGS. 1 to 9 will be omitted.
- the processor 13 may be at least one general-purpose processor.
- the processor 13 may include at least one processor manufactured to perform the function of the artificial intelligence model.
- the processor 13 may execute a series of instructions so that the artificial intelligence model learns new training data.
- the processor 13 may perform the function of the artificial intelligence model described above with reference to FIGS. 1 to 9 by executing the software module stored in the memory 17 .
- the communication unit 15 may include one or more components that allow the electronic device 10 to communicate with another device (not shown) and the server 20 .
- Another device (not shown) may be a computing device such as the electronic device 10, but is not limited thereto.
- the memory 17 may store at least one instruction and at least one program for processing and control of the processor 13 , and may store data input to or output from the electronic device 10 . have.
- the memory 17 is a memory that temporarily stores data, such as a random access memory (RAM), a static random access memory (SRAM), a flash memory type, a hard disk type, and a multimedia card.
- Multimedia card micro type card type memory (such as SD or XD memory), ROM (Read-Only Memory), EEPROM (Electrically Erasable Programmable Read-Only Memory), PROM (Programmable Read-Memory) Only memory), a magnetic memory, a magnetic disk, and an optical disk may include at least one type of storage medium among data storage for non-temporarily storing data.
- FIG. 11 is a block diagram of a software module stored in a memory of an electronic device, according to an exemplary embodiment.
- the memory 17 is a software module including instructions for the electronic device 10 to perform the embodiments described above with reference to FIGS. 1 to 9 , and includes a face identification and face recognition module ( 17a), a facial feature acquisition module 17b, an image quality improvement module 17c, and an artificial intelligence model learning module 17d.
- the electronic device 10 may perform image quality improvement by using more software modules than the software modules shown in FIG. 11 , and the electronic device 10 may use fewer software modules than the software modules shown in FIG. 10 . Image quality can be improved.
- the electronic device 10 may detect a face from the image and identify a person corresponding to the detected face. .
- the content overlapping with the embodiment described above with reference to FIGS. 1 to 9 will be omitted.
- the electronic device 10 may acquire the facial features detected from the image by the processor 13 executing the instructions stored in the facial feature acquiring module 17b.
- the content overlapping with the embodiment described above with reference to FIGS. 1 to 9 will be omitted.
- the electronic device 10 may improve the image quality of the image.
- the content overlapping with the embodiment described above with reference to FIGS. 1 to 9 will be omitted.
- the electronic device 10 may train the artificial intelligence model by executing the instructions stored in the artificial intelligence model learning module 17d by the processor 13 .
- the content overlapping with the embodiment described above with reference to FIGS. 1 to 9 will be omitted.
- FIG. 12 is a block diagram of a server, according to an embodiment.
- the server 20 may perform at least one operation of the electronic device 10 .
- the server 20 may perform at least one operation among the operations of the artificial intelligence model 19 described above.
- the server 20 may include a communication unit 25 , a memory 26 , a DB 27 , and a processor 23 .
- the communication unit 25 may include one or more components that allow the server 20 to communicate with the electronic device 10 .
- the memory 26 may store at least one instruction and at least one program for processing and control of the processor 23 , and may store data input to or output from the server 20 .
- the DB 27 may store data received from the electronic device 10 .
- the DB 27 may store a plurality of training data sets to be used for learning the artificial intelligence model.
- the processor 23 typically controls the overall operation of the server 20 .
- the processor 23 may control the DB 27 and the communication unit 25 in general by executing programs stored in the memory 26 of the server 20 .
- the processor 23 may perform at least one of the operations of the electronic device 10 described with reference to FIGS. 1 to 9 and the operations of the server 20 .
- the processor 23 may control the server 20 to detect a face from an image and identify a person corresponding to the detected face by executing instructions stored in the face identification and face recognition module 17a. have.
- the content overlapping with the embodiment described above with reference to FIGS. 1 to 9 will be omitted.
- the processor 23 may control the server 20 to acquire the facial features detected from the image by executing the instructions stored in the facial feature acquisition module 17b.
- the content overlapping with the embodiment described above with reference to FIGS. 1 to 9 will be omitted.
- the processor 23 may control the server 20 to improve the image quality by executing an instruction stored in the image quality improvement module 17c.
- the content overlapping with the embodiment described above with reference to FIGS. 1 to 9 will be omitted.
- the processor 23 may control the server 20 so that the artificial intelligence model learns the training data by executing the instructions stored in the artificial intelligence model learning module 17d.
- the content overlapping with the embodiment described above with reference to FIGS. 1 to 9 will be omitted.
- FIG. 13 is a block diagram of a software module stored in a memory of a server, according to an embodiment.
- the memory 26 is a software module for the server 20 to perform the embodiments described above with reference to FIGS. 1 to 9 , a face identification and face recognition module 17a , a facial feature acquisition module (17b), an image quality improvement module 17c, and an artificial intelligence model learning module 17d.
- the server 20 can perform image quality improvement by more software modules than the software modules shown in FIG. 13, and the server 20 can improve the image quality by using fewer software modules than the software modules shown in FIG. Image quality can be improved.
- the electronic device 10 may detect a face from the image and identify a person corresponding to the detected face. .
- the content overlapping with the embodiment described above with reference to FIGS. 1 to 9 will be omitted.
- the electronic device 10 may acquire the facial features detected from the image by the processor 13 executing the instructions stored in the facial feature acquiring module 17b.
- the content overlapping with the embodiment described above with reference to FIGS. 1 to 9 will be omitted.
- the electronic device 10 may improve the image quality of the image.
- the content overlapping with the embodiment described above with reference to FIGS. 1 to 9 will be omitted.
- the electronic device 10 may train the artificial intelligence model by executing the instructions stored in the artificial intelligence model learning module 17d by the processor 13 .
- the content overlapping with the embodiment described above with reference to FIGS. 1 to 9 will be omitted.
- Computer-readable media can be any available media that can be accessed by a computer and includes both volatile and nonvolatile media, removable and non-removable media. Also, computer-readable media may include computer storage media. Computer storage media includes both volatile and nonvolatile, removable and non-removable media implemented in any method or technology for storage of information such as computer readable instructions, data structures, program modules or other data.
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Health & Medical Sciences (AREA)
- Oral & Maxillofacial Surgery (AREA)
- General Health & Medical Sciences (AREA)
- Human Computer Interaction (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- Signal Processing (AREA)
- Artificial Intelligence (AREA)
- Quality & Reliability (AREA)
- Computing Systems (AREA)
- Databases & Information Systems (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
- Measuring Pulse, Heart Rate, Blood Pressure Or Blood Flow (AREA)
- Measuring And Recording Apparatus For Diagnosis (AREA)
Abstract
Description
Claims (15)
- 전자 장치가 인공지능 모델을 이용하여 저화질의 제1 인물 이미지에 대해서 이미지 프로세싱을 수행함으로써, 고화질의 제2 인물 이미지를 생성하는 방법에 있어서,상기 제1 인물 이미지를 식별하는 단계;상기 제1 인물 이미지를 인공지능 모델에 적용하는 단계; 및상기 인공지능 모델로부터 출력되는 상기 제2 인물 이미지를 획득하는 단계를 포함하고,상기 인공지능 모델은,안면 인식 인공지능 모델을 이용하여, 상기 제1 인물 이미지에 대해서 안면 식별 및 안면 인식을 수행함으로써, 제1 안면을 인식하고,화질 개선 인공지능 모델을 이용하여, 상기 제1 안면에 대응하는 영역에 대해서 화질을 개선하는 이미지 프로세싱을 수행함으로써, 제2 인물 이미지를 획득하고,상기 제2 인물 이미지를 상기 전자 장치로 출력하는,방법.
- 제1항에 있어서,상기 인공지능 모델은,복수의 고화질 인물 이미지들의 각각이 변환된 복수의 저화질 인물 이미지 및 상기 복수의 고화질 인물 이미지를 학습 데이터로서 학습함으로써, 상기 화질 개선 인공지능 모델을 갱신하고,상기 갱신된 화질 개선 인공지능 모델을 이용하여, 상기 제1 인물 이미지로부터 상기 제2 인물 이미지를 획득하는,방법.
- 제2항에 있어서,상기 인공지능 모델은,제1 고화질 인물 이미지와 상기 제1 고화질 인물 이미지에 대해서 이미지 열화가 적용됨으로써 생성된 제1 저화질 인물 이미지를 쌍으로 학습함으로써, 상기 제1 저화질 인물 이미지로부터 상기 제1 고화질 인물 이미지를 생성하도록 상기 화질 개선 인공지능 모델을 갱신하고,상기 갱신된 화질 개선 인공지능 모델을 이용하여, 상기 제1 인물 이미지로부터 상기 제2 인물 이미지를 획득하는,방법.
- 제1항에 있어서,상기 인공지능 모델은,인물별로 분류된 복수의 인물 이미지들을 이용하여 개인화 학습을 함으로써, 안면 인식 인공지능 모델 및 화질 개선 인공지능 모델을 갱신하고,상기 갱신된 안면 인식 인공지능 모델을 이용하여, 상기 제1 인물 이미지로부터 상기 제1 안면 및 상기 제1 안면에 대응하는 제1 인물을 식별하고,상기 제1 인물에 대하여 갱신된 화질 개선 인공지능 모델을 이용하여, 상기 제1 인물 이미지로부터 상기 제2 인물 이미지를 획득하는,방법.
- 제4항에 있어서,상기 인공지능 모델은,상기 개인화 학습으로 갱신된 상기 안면 인식 인공지능 모델 및 화질 개선 인공지능 모델에 대해서 경량화를 수행하고,상기 경량화된 안면 인식 인공지능 모델을 이용하여, 상기 제1 인물 이미지로부터 상기 제1 인물을 식별하고,상기 경량화된 화질 개선 인공지능 모델을 이용하여, 상기 제1 인물 이미지로부터 상기 제1 인물에 대해서 화질을 개선한 상기 제2 인물 이미지를 획득하는,방법.
- 제4항에 있어서,상기 인공지능 모델은,상기 제1 인물에 대한 복수의 인물 이미지를 학습 데이터로서 학습함으로써, 상기 제1 인물의 안면 특징을 획득하고,상기 제1 인물의 안면 특징에 기초하여, 상기 제1 인물 이미지로부터 상기 제1 인물을 식별하고,상기 제1 인물의 안면 특징에 기초하여, 상기 제1 인물 이미지로부터 상기 제1 인물의 화질을 개선한 상기 제2 인물 이미지를 획득하는,방법.
- 제6항에 있어서,상기 방법은,사용자로부터 상기 제1 인물의 화질 개선 부위를 선택하는 입력을 수신하는 단계;상기 인공지능 모델로 상기 사용자가 선택한 화질 개선 부위에 대한 정보를 적용하는 단계를 더 포함하고,상기 인공지능 모델은,상기 제1 인물의 상기 화질 개선 부위에 대한 학습 데이터를 추가 학습함으로써, 상기 화질 개선 인공지능 모델을 갱신하고,상기 갱신된 화질 개선 인공지능 모델을 이용하여, 상기 제1 인물의 상기 제1 안면에 대응하는 영역 중에서 상기 사용자가 선택한 화질 개선 부위에 대응하는 영역의 화질이 개선된 상기 제2 인물 이미지를 획득하는,방법.
- 제6항에 있어서,상기 방법은,사용자로부터 상기 제1 인물에 대한 화질 개선 방향을 선택하는 입력을 수신하는 단계;상기 인공지능 모델로 상기 사용자가 선택한 화질 개선 방향에 대한 정보를 적용하는 단계를 더 포함하고,상기 인공지능 모델은,상기 화질 개선 방향에 대한 학습 데이터를 추가 학습함으로써, 상기 화질 개선 인공지능 모델을 갱신하고,상기 갱신된 화질 개선 인공지능 모델을 이용하여, 상기 화질 개선 방향에 따라서 상기 제1 인물의 안면 특징을 수정함으로써 상기 제2 인물 이미지를 획득하는, 방법.
- 제8항에 있어서,상기 방법은,제2 인물을 지정하는 사용자 입력을 수신하는 단계;상기 제2 인물에 대한 데이터를 획득하는 단계;상기 제2 인물에 대한 데이터를 학습 데이터로서 인공지능 모델에 적용하는 단계;상기 인공지능 모델은,상기 제2 인물에 대한 데이터로부터 상기 제2 인물의 안면 특징을 획득하고,상기 제2 인물의 안면 특징에 기초하여, 상기 제1 인물의 안면 특징을 수정함으로써 상기 제2 인물 이미지를 획득하는,방법.
- 인공지능 모델을 이용하여 저화질의 제1 인물 이미지에 대해서 이미지 프로세싱을 수행함으로써, 고화질의 제2 인물 이미지를 생성하는 전자 장치가 읽을 수 있는 적어도 하나의 명령어들이 저장된 기록 매체에 있어서, 상기 기록 매체는, 상기 전자 장치가 상기 적어도 하나의 명령어를 실행함으로써,상기 제1 인물 이미지를 식별하고,상기 제1 인물 이미지를 인공지능 모델에 적용하고,상기 인공지능 모델로부터 출력되는 상기 제2 인물 이미지를 획득하고,상기 인공지능 모델은,안면 인식 인공지능 모델을 이용하여, 상기 제1 인물 이미지에 대해서 안면 식별 및 안면 인식을 수행함으로써, 제1 안면을 인식하고,화질 개선 인공지능 모델을 이용하여, 상기 제1 안면에 대응하는 영역에 대해서 화질을 개선하는 이미지 프로세싱을 수행함으로써, 제2 인물 이미지를 획득하고,상기 제2 인물 이미지를 상기 전자 장치로 출력하는,기록 매체.
- 인공지능 모델을 이용하여 저화질의 제1 인물 이미지에 대해서 이미지 프로세싱을 수행함으로써, 고화질의 제2 인물 이미지를 생성하는 전자 장치에 있어서,적어도 하나의 명령어를 저장하는 메모리; 및상기 적어도 하나의 명령어를 실행하는 프로세서를 포함하고,상기 프로세서는 상기 적어도 하나의 명령어를 실행함으로써,상기 제1 인물 이미지를 식별하고,상기 제1 인물 이미지를 인공지능 모델에 적용하고,상기 인공지능 모델로부터 출력되는 상기 제2 인물 이미지를 획득하고,상기 인공지능 모델은,안면 인식 인공지능 모델을 이용하여, 상기 제1 인물 이미지에 대해서 안면 식별 및 안면 인식을 수행함으로써, 제1 안면을 인식하고,화질 개선 인공지능 모델을 이용하여, 상기 제1 안면에 대응하는 영역에 대해서 화질을 개선하는 이미지 프로세싱을 수행함으로써, 제2 인물 이미지를 획득하고,상기 제2 인물 이미지를 상기 전자 장치로 출력하는,전자 장치.
- 제11항에 있어서,상기 인공지능 모델은,복수의 고화질 인물 이미지들의 각각이 변환된 복수의 저화질 인물 이미지 및 상기 복수의 고화질 인물 이미지를 학습 데이터로서 학습함으로써, 상기 화질 개선 인공지능 모델을 갱신하고,상기 갱신된 화질 개선 인공지능 모델을 이용하여, 상기 제1 인물 이미지로부터 상기 제2 인물 이미지를 획득하는,전자 장치.
- 제12항에 있어서,상기 인공지능 모델은,제1 고화질 인물 이미지와 상기 제1 고화질 인물 이미지에 대해서 이미지 열화가 적용됨으로써 생성된 제1 저화질 인물 이미지를 쌍으로 학습함으로써, 상기 제1 저화질 인물 이미지로부터 상기 제1 고화질 인물 이미지를 생성하도록 상기 화질 개선 인공지능 모델을 갱신하고,상기 갱신된 화질 개선 인공지능 모델을 이용하여, 상기 제1 인물 이미지로부터 상기 제2 인물 이미지를 획득하는,전자 장치.
- 제11항에 있어서,상기 인공지능 모델은,인물별로 분류된 복수의 인물 이미지들을 이용하여 개인화 학습함으로써, 안면 인식 인공지능 모델 및 화질 개선 인공지능 모델을 갱신하고,상기 갱신된 안면 인식 인공지능 모델을 이용하여, 상기 제1 인물 이미지로부터 상기 제1 안면 및 상기 제1 안면에 대응하는 제1 인물을 식별하고,상기 제1 인물에 대하여 갱신된 화질 개선 인공지능 모델을 이용하여, 상기 제1 인물 이미지로부터 상기 제2 인물 이미지를 획득하는,전자 장치.
- 제14항에 있어서,상기 인공지능 모델은,상기 개인화 학습으로 갱신된 상기 안면 인식 인공지능 모델 및 화질 개선 인공지능 모델에 대해서 경량화를 수행하고,상기 경량화된 안면 인식 인공지능 모델을 이용하여, 상기 제1 인물 이미지로부터 상기 제1 인물을 식별하고,상기 경량화된 화질 개선 인공지능 모델을 이용하여, 상기 제1 인물 이미지로부터 상기 제1 인물에 대해서 화질을 개선한 상기 제2 인물 이미지를 획득하는,전자 장치.
Priority Applications (3)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
EP22767370.4A EP4303805A1 (en) | 2021-03-09 | 2022-02-23 | Electronic device for improving quality of image and method for improving quality of image by using same |
CN202280019697.XA CN116964619A (zh) | 2021-03-09 | 2022-02-23 | 用于增强图像质量的电子设备以及通过使用该电子设备增强图像质量的方法 |
US18/244,088 US20230419721A1 (en) | 2021-03-09 | 2023-09-08 | Electronic device for improving quality of image and method for improving quality of image by using same |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
KR10-2021-0030934 | 2021-03-09 | ||
KR1020210030934A KR20220126535A (ko) | 2021-03-09 | 2021-03-09 | 이미지의 화질을 개선하는 전자 장치 및 이를 이용한 이미지의 화질 개선 방법 |
Related Child Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
US18/244,088 Continuation US20230419721A1 (en) | 2021-03-09 | 2023-09-08 | Electronic device for improving quality of image and method for improving quality of image by using same |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2022191474A1 true WO2022191474A1 (ko) | 2022-09-15 |
Family
ID=83226884
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/KR2022/002649 WO2022191474A1 (ko) | 2021-03-09 | 2022-02-23 | 이미지의 화질을 개선하는 전자 장치 및 이를 이용한 이미지의 화질 개선 방법 |
Country Status (5)
Country | Link |
---|---|
US (1) | US20230419721A1 (ko) |
EP (1) | EP4303805A1 (ko) |
KR (1) | KR20220126535A (ko) |
CN (1) | CN116964619A (ko) |
WO (1) | WO2022191474A1 (ko) |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2024096662A1 (ko) * | 2022-11-04 | 2024-05-10 | 삼성전자 주식회사 | 투명 배경 이미지의 화질을 분석 및 개선하는 전자 장치, 동작 방법, 및 저장 매체 |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR20140061028A (ko) * | 2012-11-13 | 2014-05-21 | 재단법인대구경북과학기술원 | 주파수 성분과 영역 분할을 이용한 고해상도 얼굴 영상 복원 장치 및 방법 |
KR101737619B1 (ko) * | 2016-11-30 | 2017-05-19 | 윈스로드(주) | 얼굴 인식 장치 및 방법 |
JP2019501454A (ja) * | 2016-03-03 | 2019-01-17 | 三菱電機株式会社 | 画像をアップサンプリングするコンピューターシステム及び方法 |
KR20200075063A (ko) * | 2018-12-07 | 2020-06-26 | 주식회사 포스코아이씨티 | 딥러닝 기반의 얼굴이미지 추출장치 |
KR102132690B1 (ko) * | 2019-01-30 | 2020-07-13 | 인천대학교 산학협력단 | 초고해상도 영상 복원 시스템 |
-
2021
- 2021-03-09 KR KR1020210030934A patent/KR20220126535A/ko active Search and Examination
-
2022
- 2022-02-23 CN CN202280019697.XA patent/CN116964619A/zh active Pending
- 2022-02-23 WO PCT/KR2022/002649 patent/WO2022191474A1/ko active Application Filing
- 2022-02-23 EP EP22767370.4A patent/EP4303805A1/en active Pending
-
2023
- 2023-09-08 US US18/244,088 patent/US20230419721A1/en active Pending
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR20140061028A (ko) * | 2012-11-13 | 2014-05-21 | 재단법인대구경북과학기술원 | 주파수 성분과 영역 분할을 이용한 고해상도 얼굴 영상 복원 장치 및 방법 |
JP2019501454A (ja) * | 2016-03-03 | 2019-01-17 | 三菱電機株式会社 | 画像をアップサンプリングするコンピューターシステム及び方法 |
KR101737619B1 (ko) * | 2016-11-30 | 2017-05-19 | 윈스로드(주) | 얼굴 인식 장치 및 방법 |
KR20200075063A (ko) * | 2018-12-07 | 2020-06-26 | 주식회사 포스코아이씨티 | 딥러닝 기반의 얼굴이미지 추출장치 |
KR102132690B1 (ko) * | 2019-01-30 | 2020-07-13 | 인천대학교 산학협력단 | 초고해상도 영상 복원 시스템 |
Also Published As
Publication number | Publication date |
---|---|
KR20220126535A (ko) | 2022-09-16 |
EP4303805A1 (en) | 2024-01-10 |
US20230419721A1 (en) | 2023-12-28 |
CN116964619A (zh) | 2023-10-27 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2020130747A1 (ko) | 스타일 변환을 위한 영상 처리 장치 및 방법 | |
WO2018117428A1 (en) | Method and apparatus for filtering video | |
WO2020085694A1 (ko) | 이미지 획득 장치 및 그의 제어 방법 | |
WO2019143227A1 (en) | Electronic device providing text-related image and method for operating the same | |
WO2018143630A1 (ko) | 상품을 추천하는 디바이스 및 방법 | |
WO2020067633A1 (en) | Electronic device and method of obtaining emotion information | |
EP3539056A1 (en) | Electronic apparatus and operation method thereof | |
WO2019059505A1 (ko) | 객체를 인식하는 방법 및 장치 | |
WO2019027141A1 (en) | ELECTRONIC DEVICE AND METHOD FOR CONTROLLING THE OPERATION OF A VEHICLE | |
WO2020179995A1 (ko) | 전자 장치 및 이의 제어 방법 | |
WO2021112631A1 (en) | Device, method, and program for enhancing output content through iterative generation | |
WO2018117538A1 (ko) | 차선 정보를 추정하는 방법 및 전자 장치 | |
WO2021006482A1 (en) | Apparatus and method for generating image | |
WO2019050297A1 (ko) | 뉴럴 네트워크 학습 방법 및 장치 | |
WO2021150033A1 (en) | Electronic device and controlling method of electronic device | |
EP3820369A1 (en) | Electronic device and method of obtaining emotion information | |
WO2020159241A1 (ko) | 이미지를 처리하기 위한 방법 및 그에 따른 장치 | |
WO2022191474A1 (ko) | 이미지의 화질을 개선하는 전자 장치 및 이를 이용한 이미지의 화질 개선 방법 | |
WO2019107674A1 (en) | Computing apparatus and information input method of the computing apparatus | |
WO2021040192A1 (ko) | 인공지능 모델을 학습시키는 시스템 및 방법 | |
EP3545685A1 (en) | Method and apparatus for filtering video | |
WO2019190142A1 (en) | Method and device for processing image | |
EP3707678A1 (en) | Method and device for processing image | |
WO2019231068A1 (en) | Electronic device and control method thereof | |
WO2019088338A1 (ko) | 전자 장치 및 그 제어 방법 |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 22767370 Country of ref document: EP Kind code of ref document: A1 |
|
WWE | Wipo information: entry into national phase |
Ref document number: 202280019697.X Country of ref document: CN |
|
WWE | Wipo information: entry into national phase |
Ref document number: 2022767370 Country of ref document: EP |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
ENP | Entry into the national phase |
Ref document number: 2022767370 Country of ref document: EP Effective date: 20231006 |