CN114170668A - Hyperspectral face recognition method and system - Google Patents

Hyperspectral face recognition method and system Download PDF

Info

Publication number
CN114170668A
CN114170668A CN202111534924.6A CN202111534924A CN114170668A CN 114170668 A CN114170668 A CN 114170668A CN 202111534924 A CN202111534924 A CN 202111534924A CN 114170668 A CN114170668 A CN 114170668A
Authority
CN
China
Prior art keywords
image
hyperspectral
face
spectral
data
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202111534924.6A
Other languages
Chinese (zh)
Inventor
潘冬宁
武斌
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Photonic Integration Wenzhou Innovation Research Institute
Original Assignee
Photonic Integration Wenzhou Innovation Research Institute
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Photonic Integration Wenzhou Innovation Research Institute filed Critical Photonic Integration Wenzhou Innovation Research Institute
Priority to CN202111534924.6A priority Critical patent/CN114170668A/en
Publication of CN114170668A publication Critical patent/CN114170668A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
    • G06F18/2135—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods based on approximation criteria, e.g. principal component analysis
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/25—Fusion techniques
    • G06F18/253—Fusion techniques of extracted features

Landscapes

  • Engineering & Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Artificial Intelligence (AREA)
  • Evolutionary Biology (AREA)
  • Evolutionary Computation (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Image Analysis (AREA)
  • Image Processing (AREA)

Abstract

The invention discloses a hyperspectral face recognition method and system, belongs to the technical field of public safety, and can solve the problems that the face recognition precision is poor in a complex illumination environment, and an existing face recognition system does not have an anti-counterfeiting function. The method comprises the following steps: acquiring a hyperspectral image and a high-resolution RGB image of a human face; extracting a characteristic spectral band image containing human face biological characteristics in the hyperspectral image to obtain an accurate human face area in the hyperspectral image; carrying out feature positioning and edge contour feature point correction on the accurate face area to obtain a corrected image; fusing the high-resolution RGB image and the corrected image to obtain a fused image; and comparing the spectral characteristic data with data in a human face anti-counterfeiting database, and outputting a comparison result.

Description

Hyperspectral face recognition method and system
Technical Field
The invention relates to a hyperspectral face recognition method and system, and belongs to the technical field of public safety.
Background
The face recognition technology is one of the most representative technologies in the field of artificial intelligence, and identity recognition is completed mainly by extracting feature information of a face and matching the feature information with the face in a database. The face recognition technology is widely applied to the aspects of security, entrance guard, finance, new retail industry and the like in the current social life.
The existing face recognition system has high requirements on illumination conditions and human-computer directions, can only collect light intensity information, and has poor face recognition accuracy and no anti-counterfeiting function in a complex illumination environment such as a strong light, a weak light or a shadow.
Disclosure of Invention
The invention provides a hyperspectral face recognition method and system, which can solve the problems that the face recognition precision is poor in a complex illumination environment and the existing face recognition system does not have an anti-counterfeiting function.
In one aspect, the invention provides a hyperspectral face recognition method, which comprises the following steps:
acquiring a hyperspectral image and a high-resolution RGB image of a human face, wherein the resolution of the high-resolution RGB image is greater than or equal to 720P;
extracting a characteristic spectral band image containing human face biological characteristics in the hyperspectral image to obtain an accurate human face area in the hyperspectral image;
carrying out feature positioning and edge contour feature point correction on the accurate face area to obtain a corrected image;
fusing the high-resolution RGB image and the corrected image to obtain a fused image;
extracting spectral characteristics of the fused image to obtain spectral characteristic data of the fused image;
comparing the spectral characteristic data with data in a human face anti-counterfeiting database, and outputting a comparison result; the data in the face anti-counterfeiting database comprises living face spectral data and spectral data of various camouflage materials.
Optionally, the spatial resolution of the fused image is greater than or equal to 1024 × 768, and the signal-to-noise ratio is greater than or equal to 60 dB.
Optionally, the spectral feature includes a comparison value of color, gray scale, or brightness between bands of a target object in an image obtained by performing point operation on each pixel point in the fused image through an original band;
the spectral features further comprise textural features;
preferably, the texture features include angular second moment, mean, variance and entropy.
Optionally, the extracting the feature spectrum image including the human face biological feature in the hyperspectral image specifically includes:
and extracting a characteristic spectral band image containing the human face biological characteristics in the hyperspectral image by a human face contour extraction method based on multi-region cooperation fitting.
Optionally, after extracting a feature spectrum image including a human face biological feature in the hyperspectral image to obtain an accurate human face region in the hyperspectral image, the method further includes:
carrying out image enhancement on the accurate face area by a self-adaptive enhancement method;
dividing the accurate face area into an under-illumination sub-area, an under-illumination sub-area and an over-exposure sub-area according to the texture gray scale distribution of the accurate face area;
optimizing the underilluminated subarea and the overexposed subarea so that the contrast of the underilluminated subarea and the overexposed subarea after optimization is larger than 100: 1.
optionally, the optimizing the under-illumination sub-region and the over-exposure sub-region specifically includes:
when the contrast of the underilluminated sub-region is less than or equal to 30: 1, processing the sub-area with insufficient illumination by a homomorphic filtering method;
when the contrast of the over-exposed subarea is less than or equal to 30: and 1, processing the overexposed subarea by using a wavelet domain-based image fusion method.
Optionally, after the fusing the high-resolution RGB image and the corrected image to obtain a fused image, the method further includes:
and denoising the fused image.
On the other hand, the invention provides a face recognition system applying the face recognition method, and the system comprises an imaging lens group, a hyperspectral camera and a processor;
the imaging lens group is used for imaging a human face;
the hyperspectral camera is used for acquiring a hyperspectral image and a high-resolution RGB image of a human face, and processing and correcting the hyperspectral image to obtain a corrected image;
the hyperspectral camera is also used for fusing the high-resolution RGB image and the corrected image to obtain a fused image, and inputting the fused image into the processor;
the processor is used for extracting spectral characteristics of the fused image to obtain spectral characteristic data of the fused image;
the processor is also used for comparing the spectral characteristic data with data in a human face anti-counterfeiting database and outputting a comparison result.
Optionally, the hyperspectral camera comprises a hyperspectral imaging sensor;
the hyperspectral imaging sensor comprises a beam splitter, a pixel filtering detector array circuit board and an RGB detector circuit board;
the beam splitter is arranged on the light path of the imaging lens group and is used for splitting the light beam into a transmission light beam and a reflection light beam;
the pixel filtering detector array circuit board and the RGB detector circuit board are symmetrically arranged on two sides of the beam splitter;
the pixel filter detector array circuit board is used for receiving the transmitted light beams to form a hyperspectral image, and the RGB detector circuit board is used for receiving the reflected light beams to form a high-resolution RGB image.
Optionally, the hyperspectral camera further comprises a hyperspectral image data interface board;
the hyperspectral image data interface board comprises an FPGA and a DSP;
the FPGA is used for sequential control, acquisition and transmission of the hyperspectral image and the high-resolution RGB image;
the DSP is used for fusion processing of the hyperspectral image and the high-resolution RGB image and transmitting the hyperspectral image data after fusion processing to the processor.
The invention can produce the beneficial effects that:
the invention utilizes the characteristic that the real face and the camouflage material have different spectral characteristics to compare the spectral characteristic data of the real face with the spectral data of various camouflage materials in the face anti-counterfeiting database, thereby realizing the anti-counterfeiting identification of the face;
the method sequentially performs image enhancement on the accurate face area, divides the accurate face area into an under-illumination subarea, a well-illuminated subarea and an over-exposure subarea according to the texture gray distribution of the accurate face area, optimizes the under-illumination subarea and the over-exposure subarea, performs feature positioning and edge contour feature point correction on the accurate face area and performs denoising processing on the fused image, and improves the face recognition precision in a complex illumination environment step by step.
Drawings
FIG. 1 is a flow chart of a method of a hyperspectral face recognition method according to an embodiment of the invention;
fig. 2 is a schematic diagram illustrating a manufacturing process of the face recognition system according to the embodiment of the present invention.
List of parts and reference numerals:
1. an imaging lens group; 2. a hyperspectral camera; 21. a beam splitter; 22. a pixel filtering detector array circuit board; 23. an RGB detector circuit board; 24. a hyperspectral image data interface board; 3. a processor.
Detailed Description
The present invention will be described in detail with reference to examples, but the present invention is not limited to these examples.
The hyperspectral images are imaged on continuous spectrum wave bands, compared with traditional RGB images, the hyperspectral images are spread in spectrum dimensions in multiple channels, and a data cube which not only has space image information, but also comprises spectrum information on any pixel point of each image is formed.
The spectrum is relative to the substance, like a fingerprint is relative to an individual, and the specific spectrum corresponds to the specific substance, so that the real human face and the camouflage material have different spectral characteristics.
On one hand, as shown in fig. 1, an embodiment of the present invention provides a hyperspectral face recognition method, including:
and S1, acquiring a hyperspectral image and a high-resolution RGB image of the human face, wherein the resolution of the high-resolution RGB image is greater than or equal to 720P, and the hyperspectral image can be one or more than one.
And S2, extracting a characteristic spectrum image containing the human face biological characteristics in the hyperspectral image to obtain an accurate human face area in the hyperspectral image.
Specifically, a feature spectral band image containing human face biological features in the hyperspectral image is extracted through a human face contour extraction method based on multi-region cooperation fitting.
Specifically, the method further comprises:
carrying out image enhancement on the accurate face area by a self-adaptive enhancement method;
dividing the accurate face area into an under-illumination sub-area, a well-illumination sub-area and an over-exposure sub-area according to the texture gray distribution of the accurate face area;
optimizing the underillumination subarea and the overexposure subarea so that the contrast of the underillumination subarea and the overexposure subarea after optimization is larger than 100: 1.
the optimizing the underilluminated subarea and the overexposed subarea specifically comprises:
when the contrast of the underilluminated subarea is less than or equal to 30: 1, the image contrast of the insufficient illumination subarea is low due to weak light and shadow conditions, and the insufficient illumination subarea is processed by a homomorphic filtering method, so that the image contrast is improved;
when the contrast of the overexposed subarea is less than or equal to 30: 1, the overexposure subarea is subjected to highlight interference caused by strong light, and the overexposure subarea is processed by a wavelet domain-based image fusion method, so that the image quality is improved.
And S3, performing feature positioning and edge contour feature point correction on the accurate face region to obtain a corrected image.
Specifically, feature positioning and edge contour feature point correction are carried out on the accurate face region through an active appearance model algorithm.
And S4, fusing the high-resolution RGB image and the corrected image to obtain a fused image.
Specifically, the high-resolution RGB image and the corrected image are fused through a self-adaptive spline enhanced interpolation algorithm, and a fused image with spatial resolution larger than or equal to 1024 × 768 and signal-to-noise ratio larger than or equal to 60dB is obtained.
The method further comprises the following steps: and denoising the fused image.
Specifically, denoising the fused image by a standard normal variation method; and carrying out image enhancement on the image subjected to denoising processing through algorithms such as filtering, sharpening, histogram equalization and the like, so as to improve the image quality.
And S5, performing spectral feature extraction on the fused image to obtain spectral feature data of the fused image.
Specifically, the spectral feature extraction is carried out on the fused image by a principal component analysis method and a competitive adaptive weighting method;
establishing a machine algorithm identification model by using the extracted spectral characteristics as input through a least square method and a support vector machine method;
taking the characteristic data of the training data set as the input parameters of the machine algorithm recognition model, and training the machine algorithm recognition model by combining the known result corresponding to the training data set;
after training is finished, the machine algorithm recognition model can predict a processing result according to the feature data of the verification data set, namely the machine algorithm recognition model can obtain corresponding spectral feature data according to the input spectral features.
The spectral characteristics comprise comparison values of the color, the gray level or the brightness between wave bands of a target object in an image obtained by point operation of an original wave band of each pixel point in the fused image;
the spectral features also include texture features such as angular second moment, mean, variance, entropy, and the like.
S6, comparing the spectral feature data with data in a face anti-counterfeiting database, and outputting a comparison result; the data in the face anti-counterfeiting database comprises living face spectral data and spectral data of various camouflage materials.
On the other hand, as shown in fig. 2, the present invention provides a face recognition system applying the face recognition method, and the system includes an imaging lens group 1, a hyperspectral camera 2 and a processor 5.
The imaging lens group 1 is composed of a plurality of optical lenses and is used for imaging a human face.
The hyperspectral camera 2 is used for acquiring a hyperspectral image and a high-resolution RGB image of a human face, and processing and correcting the hyperspectral image to obtain a corrected image;
the hyperspectral camera 2 is further configured to fuse the high-resolution RGB image and the corrected image to obtain a fused image, and input the fused image to the processor 5.
The processor 5 is used for extracting spectral characteristics of the fused image to obtain spectral characteristic data of the fused image;
the processor 5 is also used for comparing the spectral characteristic data with the data in the human face anti-counterfeiting database and outputting a comparison result.
Specifically, the hyperspectral camera 2 includes a hyperspectral imaging sensor and a hyperspectral image data interface board 24.
The hyperspectral imaging sensor comprises a beam splitter 21, a pixel filtering detector array circuit board 22 and an RGB detector circuit board 23;
the beam splitter 21 is arranged on the light path of the imaging lens group 1 and is used for splitting the light beam into a transmitted light beam and a reflected light beam;
the pixel filtering detector array circuit board 22 and the RGB detector circuit board 23 are symmetrically arranged at two sides of the beam splitter 21;
the pixel filter detector array circuit board 22 includes a pixel filter detector array and a driving circuit thereof, and is configured to receive the transmitted light beams to form a hyperspectral image.
In this embodiment, the pixel filtering detector array is a mature snapshot type mosaic spectrum sensor of the IMEC, and has a spectrum band number of 25, a spectrum bandwidth of about 10nm, and a spatial resolution of about 409 × 216.
The RGB detector circuit board 23 includes a high spatial resolution RGB detector and its driving circuitry for receiving the reflected beam to form a high resolution RGB image having a resolution greater than or equal to 720P and a pixel filter detector array of the same size as the high spatial resolution RGB detector.
The hyperspectral image data interface board 24 comprises an FPGA and a DSP, the FPGA is used for sequential control, acquisition and transmission of the hyperspectral image and the high-resolution RGB image, and the DSP is used for fusion processing of the hyperspectral image and the high-resolution RGB image and transmitting the hyperspectral image data after the fusion processing to the processor 5 through a camera link or lvds interface.
The processor 5 is a computer.
Although the present application has been described with reference to a preferred embodiment, it should be understood that various changes, substitutions and alterations can be made herein without departing from the spirit and scope of the application.

Claims (10)

1. A hyperspectral face recognition method is characterized by comprising the following steps:
acquiring a hyperspectral image and a high-resolution RGB image of a human face, wherein the resolution of the high-resolution RGB image is greater than or equal to 720P;
extracting a characteristic spectral band image containing human face biological characteristics in the hyperspectral image to obtain an accurate human face area in the hyperspectral image;
carrying out feature positioning and edge contour feature point correction on the accurate face area to obtain a corrected image;
fusing the high-resolution RGB image and the corrected image to obtain a fused image;
extracting spectral characteristics of the fused image to obtain spectral characteristic data of the fused image;
comparing the spectral characteristic data with data in a human face anti-counterfeiting database, and outputting a comparison result; the data in the face anti-counterfeiting database comprises living face spectral data and spectral data of various camouflage materials.
2. The method of claim 1, wherein the fused image has a spatial resolution greater than or equal to 1024 x 768 and a signal-to-noise ratio greater than or equal to 60 dB.
3. The face recognition method according to claim 1, wherein the spectral features include comparison values of color, gray scale or brightness between bands of a target object in an image obtained by point operation of an original band for each pixel point in the fused image;
the spectral features further comprise textural features;
preferably, the texture features include angular second moment, mean, variance and entropy.
4. The face recognition method according to claim 1, wherein the extracting of the feature spectrum image including the face biometric feature from the hyperspectral image specifically comprises:
and extracting a characteristic spectral band image containing the human face biological characteristics in the hyperspectral image by a human face contour extraction method based on multi-region cooperation fitting.
5. The face recognition method according to claim 1, wherein after the extracting a feature spectrum image containing a face biological feature from the hyperspectral image to obtain an accurate face region in the hyperspectral image, the method further comprises:
carrying out image enhancement on the accurate face area by a self-adaptive enhancement method;
dividing the accurate face area into an under-illumination sub-area, an under-illumination sub-area and an over-exposure sub-area according to the texture gray scale distribution of the accurate face area;
optimizing the underilluminated subarea and the overexposed subarea so that the contrast of the underilluminated subarea and the overexposed subarea after optimization is larger than 100: 1.
6. the face recognition method of claim 5, wherein the optimizing the under-illuminated sub-region and the over-exposed sub-region specifically comprises:
when the contrast of the underilluminated sub-region is less than or equal to 30: 1, processing the sub-area with insufficient illumination by a homomorphic filtering method;
when the contrast of the over-exposed subarea is less than or equal to 30: and 1, processing the overexposed subarea by using a wavelet domain-based image fusion method.
7. The method of claim 1, wherein after the fusing the RGB images with the corrected images to obtain fused images, the method further comprises:
and denoising the fused image.
8. A face recognition system applying the face recognition method according to any one of claims 1 to 7, wherein the system comprises an imaging lens group, a hyperspectral camera and a processor;
the imaging lens group is used for imaging a human face;
the hyperspectral camera is used for acquiring a hyperspectral image and a high-resolution RGB image of a human face, and processing and correcting the hyperspectral image to obtain a corrected image;
the hyperspectral camera is also used for fusing the high-resolution RGB image and the corrected image to obtain a fused image, and inputting the fused image into the processor;
the processor is used for extracting spectral characteristics of the fused image to obtain spectral characteristic data of the fused image;
the processor is also used for comparing the spectral characteristic data with data in a human face anti-counterfeiting database and outputting a comparison result.
9. The face recognition system of claim 8, wherein the hyperspectral camera comprises a hyperspectral imaging sensor;
the hyperspectral imaging sensor comprises a beam splitter, a pixel filtering detector array circuit board and an RGB detector circuit board;
the beam splitter is arranged on the light path of the imaging lens group and is used for splitting the light beam into a transmission light beam and a reflection light beam;
the pixel filtering detector array circuit board and the RGB detector circuit board are symmetrically arranged on two sides of the beam splitter;
the pixel filter detector array circuit board is used for receiving the transmitted light beams to form a hyperspectral image, and the RGB detector circuit board is used for receiving the reflected light beams to form a high-resolution RGB image.
10. The face recognition system of claim 8, wherein the hyperspectral camera further comprises a hyperspectral image data interface board;
the hyperspectral image data interface board comprises an FPGA and a DSP;
the FPGA is used for sequential control, acquisition and transmission of the hyperspectral image and the high-resolution RGB image;
the DSP is used for fusion processing of the hyperspectral image and the high-resolution RGB image and transmitting the hyperspectral image data after fusion processing to the processor.
CN202111534924.6A 2021-12-15 2021-12-15 Hyperspectral face recognition method and system Pending CN114170668A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202111534924.6A CN114170668A (en) 2021-12-15 2021-12-15 Hyperspectral face recognition method and system

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202111534924.6A CN114170668A (en) 2021-12-15 2021-12-15 Hyperspectral face recognition method and system

Publications (1)

Publication Number Publication Date
CN114170668A true CN114170668A (en) 2022-03-11

Family

ID=80486724

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202111534924.6A Pending CN114170668A (en) 2021-12-15 2021-12-15 Hyperspectral face recognition method and system

Country Status (1)

Country Link
CN (1) CN114170668A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2024066118A1 (en) * 2022-09-27 2024-04-04 同济大学 Plastic debris and microplastics detection method based on rgb and hyperspectral image fusion

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2024066118A1 (en) * 2022-09-27 2024-04-04 同济大学 Plastic debris and microplastics detection method based on rgb and hyperspectral image fusion

Similar Documents

Publication Publication Date Title
Wang et al. Semantic segmentation of crop and weed using an encoder-decoder network and image enhancement method under uncontrolled outdoor illumination
CN108197546B (en) Illumination processing method and device in face recognition, computer equipment and storage medium
EP0932114B1 (en) A method of and apparatus for detecting a face-like region
CN111046703A (en) Face anti-counterfeiting detection method and device and multi-view camera
CN111462128B (en) Pixel-level image segmentation system and method based on multi-mode spectrum image
CN109087315B (en) Image identification and positioning method based on convolutional neural network
WO2016175234A1 (en) Color image processing method, color image processing program, and object recognition method and device
JP6497579B2 (en) Image composition system, image composition method, image composition program
CN112258579A (en) Image processing method, image processing device, electronic equipment and computer readable storage medium
Pouli et al. Image Statistics and their Applications in Computer Graphics.
CN109492714A (en) Image processing apparatus and its method
CN110047059B (en) Image processing method and device, electronic equipment and readable storage medium
CN109308702A (en) A kind of real-time recognition positioning method of target
CN114331937B (en) Multi-source image fusion method based on feedback iterative adjustment under low illumination condition
CN114170668A (en) Hyperspectral face recognition method and system
CN115578304B (en) Multi-band image fusion method and system combining saliency region detection
EP4167134A1 (en) System and method for maximizing inference accuracy using recaptured datasets
WO2022262760A1 (en) Micro-optical device-based end-to-end photoelectric detection system and method
CN110197178A (en) A kind of rice type of TuPu method fusion depth network quickly identifies detection device and its detection method
CN111080560B (en) Image processing and identifying method
CN111881924B (en) Dark-light vehicle illumination identification method combining illumination invariance and short-exposure illumination enhancement
CN114549386A (en) Multi-exposure image fusion method based on self-adaptive illumination consistency
CN113379611B (en) Image processing model generation method, processing method, storage medium and terminal
Zheng A channel-based color fusion technique using multispectral images for night vision enhancement
Nair et al. Benchmarking single image dehazing methods

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination