CN114494691A - Image processing method and image processing system - Google Patents

Image processing method and image processing system Download PDF

Info

Publication number
CN114494691A
CN114494691A CN202011265432.7A CN202011265432A CN114494691A CN 114494691 A CN114494691 A CN 114494691A CN 202011265432 A CN202011265432 A CN 202011265432A CN 114494691 A CN114494691 A CN 114494691A
Authority
CN
China
Prior art keywords
matrix
image
image matrix
image processing
classified
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
CN202011265432.7A
Other languages
Chinese (zh)
Inventor
陈彦萤
欧欣颖
余家伟
谢俊兴
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.)
Realtek Semiconductor Corp
Original Assignee
Realtek Semiconductor Corp
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 Realtek Semiconductor Corp filed Critical Realtek Semiconductor Corp
Priority to CN202011265432.7A priority Critical patent/CN114494691A/en
Publication of CN114494691A publication Critical patent/CN114494691A/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
    • G06F18/24—Classification techniques
    • G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00—Machine learning

Landscapes

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

Abstract

The application provides an image processing method and an image processing system. The image processing method comprises the following steps: obtaining a first image matrix; generating a first classified image matrix according to the first image matrix, wherein the first classified image matrix comprises a plurality of parts corresponding to a plurality of classes; obtaining a plurality of weights used for first image processing corresponding to a plurality of parts of the first classified image matrix, and generating a first weight matrix according to the weights; and performing the first image processing on the first image matrix according to the first weight matrix to generate a first processed image matrix.

Description

Image processing method and image processing system
Technical Field
The present disclosure relates to image processing, and more particularly, to an image processing method and an image processing system.
Background
In general image processing, by analyzing, processing and processing images, visual, psychological or other requirements of users can be met, and the viewing experience of users is improved. In the past, image processing was mostly performed by optical devices in analog mode, but due to the great increase in computer speed, these techniques are rapidly being replaced by digital image processing methods, and there is still much room for improvement.
Disclosure of Invention
The application discloses an image processing method, comprising the following steps: obtaining a first image matrix; generating a first classified image matrix according to the first image matrix, wherein the first classified image matrix comprises a plurality of parts corresponding to a plurality of classes; obtaining a plurality of weights for first image processing corresponding to a plurality of parts of the first classified image matrix, and generating a first weight matrix according to the weights; and performing the first image processing on the first image matrix according to the first weight matrix to generate a first processed image matrix.
The present application discloses an image processing system, comprising: a receiving unit for obtaining a first image matrix; a non-transitory computer readable medium storing a plurality of computer readable instructions; a processor coupled to the receiving unit and the non-transitory computer readable medium; wherein the plurality of computer readable instructions, when executed by the processor, cause the processor to: generating a first classified image matrix according to the first image matrix, wherein the first classified image matrix comprises a plurality of parts corresponding to a plurality of classes; obtaining a plurality of weights which are used for first image processing and correspond to a plurality of parts of the first classified image matrix, and generating a first weight matrix according to the weights; and an image processing unit, coupled to the receiving unit and the processor, for performing the first image processing on the first image matrix according to the first weight matrix to generate a first processed image matrix.
The image processing method and the image processing system can give different weights to a plurality of parts in the image matrix when image processing is carried out, and not carry out image processing of one view and the same kernel on the whole area of the image matrix, thereby improving the viewing experience.
Drawings
FIG. 1 is a diagram of an image processing system according to a first embodiment of the present invention.
FIG. 2 is a flowchart illustrating an embodiment of an image processing method according to the present invention.
Fig. 3 is an embodiment of an image matrix.
Fig. 4 is an embodiment of a classified image matrix.
Detailed Description
FIG. 1 is a diagram of an image processing system 100 according to a first embodiment of the present invention. The image processing system 100 includes a receiving unit 102, a non-transitory computer readable medium 104, a processor 106, and an image processing unit 108, wherein the non-transitory computer readable medium 104 stores a plurality of computer readable instructions that, when executed by the processor 106, cause the processor 106 to perform certain steps. FIG. 2 is a flowchart of an embodiment of an image processing method 200 according to the present invention. Referring to fig. 1 and fig. 2, in step 202, an image matrix I is obtained by the receiving unit 102. In the present embodiment, the image matrix I is a matrix made up of a plurality of elements. For example, where each element has a value between 0 and 255, the image matrix I may be a gray scale, red, green or blue image matrix, and a plurality of image matrices I may constitute a movie.
In step 204, the processor 106 divides the image matrix I into a plurality of portions according to the content of the image matrix I to generate a classified image matrix Is. Please refer to fig. 3 and 4, which are the image matrix I and the classified image matrix I, respectivelysExamples of (1). In the present embodiment, the processor 106 identifies the content of the image matrix I in step 204 and performs semantic segmentation to generate the image matrix Is. As can be seen from FIG. 4, the content of the original image matrix I is in the image matrix IsThe middle of the semantic segmentation is classified as sky 402, tree 404, ground 406, and car 408, but the application is not limited to the specific algorithm of semantic segmentation, for example, a deep learning model can be used to implement semantic segmentation.
The embodiments of the present application may give different weights to the above categories when performing at least one specific image processing subsequently, instead of performing the specific image processing on the whole area, thereby generating a better viewing experience, the details of which are described later. The at least one specific image processing may include spatial noise reduction (spatial noise reduction) processing, sharpness processing, brightness processing, contrast processing, saturation processing, and the like. In the embodiment, the at least one specific image processing includes a first image processing and a second image processing performed sequentially, however, the number of the at least one specific image processing is not limited in the present application, and one or more specific image processing may be performed.
In step 206, the processor 106 is further based on the image matrix IsFor the first image processing to be performed on the image matrix I, the classification parts in (1) obtain a plurality of corresponding weights, and generate a weight matrix W according to the weightsa. The processor 106 may obtain a plurality of weights for the first image processing from a predetermined first lookup table, for example, the first lookup table at least describes the image matrix IsThe weight corresponding to each classified part in (1). For example, the first image processing may be spatial noise cancellation processing, and the contents of the first lookup table comprise:
Figure BDA0002775908660000021
Figure BDA0002775908660000031
thus, the processor 106 may generate the weight matrix W based thereonaWherein, the image matrix includes a plurality of elements corresponding to a plurality of elements of the image matrix I, in other words, each element in the image matrix I has a corresponding weight recorded in the weight matrix WaIn (1).
Next, in step 208, the image processing unit 108 processes the image according to the weight matrix WaPerforming the first image processing on the image matrix I to generate a processed image matrix IaFor example, when the first image processing is the spatial noise elimination processing, the image processing unit 108 processes the sky portion in the image matrix I (i.e. corresponding to the image matrix I)sMiddle sky 402) compared to the tree and the portion of the car (i.e., corresponding image matrix I)s Tree 404 and car 408) that would process spatial noise cancellation with heavier weight to make the sky look cleaner while retaining more details of the tree and car. While for the ground in the image matrix IFace portion (i.e. corresponding to the image matrix I)sMedium ground 406), the weight for processing spatial noise cancellation is between the sky and the tree.
In step 210, the processor 106 is further based on the image matrix IsFor the second image processing to be performed on the image matrix I, the classification parts in (1) obtain a plurality of corresponding weights, and generate a weight matrix W according to the weightsbSince the second image processing is different from the first image processing, the weight matrix WbAnd a weight matrix WaMay also be different. The processor 106 may obtain a plurality of weights for the second image processing from a predetermined second lookup table, for example, the second lookup table at least describes the image matrix IsThe weight corresponding to each classified part in (1). The second image processing may be, for example, sharpness processing. Thus, the processor 106 may generate the weight matrix W based thereonbWherein, the image matrix includes a plurality of elements corresponding to a plurality of elements of the image matrix I, in other words, each element in the image matrix I has a corresponding weight recorded in the weight matrix WbIn (1).
Next, in step 212, the image processing unit 108 processes the image according to the weight matrix WbFor the processed image matrix I after the first image processingaContinuing the second image processing to generate a processed image matrix IbAnd output.
In the second embodiment of the present application, before performing step 204, the image matrix I may be scaled down according to a predetermined ratio to generate a scaled-down image matrix IdThen, in step 204, the processor 106 is enabled to perform the process according to the reduced image matrix IdWill reduce the image matrix IdInto a plurality of parts to produce a classified image matrix IsTo speed up the computation process of the processor 106. Thus, after step 206, the processor 106 needs to obtain the weight matrix WaPerforming amplification reduction according to the preset proportion to obtain an amplified weight matrix WuaStep 208 is entered to let the image processing unit 108 based on the amplified weight matrix WuaThe first image processing is performed on the image matrix I toGenerating a processed image matrix Ia. Similarly, the amplification reduction process is also required between step 210 and step 212.
In the third embodiment of the present application, after step 206, the processor 106 may first obtain the weight matrix WaSpatial filtering is performed, and step 208 is only entered; and after step 210, the processor 106 may also apply the weight matrix WbSpatial filtering is performed before proceeding to step 212.
In the fourth embodiment of the present application, where there are a plurality of image matrices I in succession constituting a movie, after step 206, the processor 106 may first obtain two weight matrices W according to the previous image matrix I and the current image matrix IaTemporal filtering (temporal filtering) is performed, and step 208 is entered; and after step 210, the processor 106 may also apply the weight matrix WbTemporal filtering is performed before proceeding to step 212.
The second to fourth embodiments described above may be combined arbitrarily as necessary. In addition, embodiments of the image processing system 100 may vary, such as burdening the processor 106 with the work of the image processing unit 108 and removing the image processing unit 108; or by modifying the operation of the processing unit 108 to a specific circuit and removing the processing unit 108 and the non-transitory computer readable medium 104.
The foregoing description has set forth briefly the features of certain embodiments of the present application so that those skilled in the art may more fully appreciate the various aspects of the present application. Those skilled in the art should appreciate that they may readily use the present disclosure as a basis for designing or modifying other processes and structures for carrying out the same purposes and/or achieving the same advantages of the embodiments introduced herein. It should be understood that the steps mentioned in the method flow chart of the present application, except the sequence specifically described, can be performed simultaneously or partially simultaneously according to the actual requirement. In addition, the above modules or method steps can be implemented by hardware, software or firmware according to the requirement of the designer. Those skilled in the art should understand that they can still make various changes, substitutions and alterations herein without departing from the spirit and scope of the present disclosure.
Description of the reference numerals
100 image processing system
102 receiving unit
104 non-transitory computer readable medium
106 processor
108 image processing unit
200 image processing method
202 to 212, step
402 sky
404 Tree
406 ground plane
408 automobile

Claims (10)

1. An image processing method, comprising:
obtaining a first image matrix;
generating a first classified image matrix according to the first image matrix, wherein the first classified image matrix comprises a plurality of parts corresponding to a plurality of classes;
obtaining a plurality of weights for first image processing corresponding to a plurality of parts of the first classified image matrix, and generating a first weight matrix according to the weights; and
according to the first weight matrix, the first image matrix is subjected to the first image processing to generate a first processed image matrix.
2. The method of claim 1, wherein generating the first classified image matrix according to the first image matrix comprises:
the first classified image matrix is generated by performing semantic segmentation based on the first image matrix.
3. The method of claim 1, wherein obtaining weights for the first image processing corresponding to portions of the first classified image matrix comprises:
a plurality of weights for the first image processing are obtained from a look-up table corresponding to the plurality of classes.
4. The method of claim 1, further comprising:
reducing the first image matrix according to a predetermined ratio to generate a first reduced image matrix, wherein generating the first classified image matrix according to the first image matrix comprises:
according to the content of the first reduced image matrix, the first reduced image matrix is divided into a plurality of parts to generate the first classified image matrix.
5. The method of claim 4, wherein the performing the first image processing on the first image matrix according to the first weight matrix to generate the first processed image matrix comprises:
the first weight matrix is amplified according to the preset proportion to generate a first amplified weight matrix, and the first image matrix is subjected to the first image processing according to the first amplified weight matrix to generate a first processed image matrix.
6. The method of claim 1, further comprising:
the first weight matrix is spatially filtered to generate a first spatially filtered weight matrix.
7. The method of claim 6, wherein performing the first image processing on the first image matrix according to the first weight matrix to generate the first processed image matrix comprises:
and performing the first image processing on the first image matrix according to the first spatial filtered weight matrix to generate the first processed image matrix.
8. The method of claim 1, further comprising:
obtaining a second image matrix;
generating a second classified image matrix according to the second image matrix, wherein the second classified image matrix comprises a plurality of parts corresponding to a plurality of categories;
obtaining a plurality of weights for the first image processing corresponding to a plurality of parts of the second classified image matrix, and generating a second weight matrix according to the weights; and
and performing the first image processing on the second image matrix according to the second weight matrix to generate a second processed image matrix.
9. The method of claim 1, further comprising:
obtaining a plurality of weights for second image processing corresponding to a plurality of parts of the first classified image matrix, and generating a third weight matrix according to the weights; and
and performing the second image processing on the first processed image matrix according to the third weight matrix to generate a third processed image matrix.
10. An image processing system, comprising:
a receiving unit for obtaining a first image matrix;
a non-transitory computer readable medium storing a plurality of computer readable instructions;
a processor coupled to the receiving unit and the non-transitory computer readable medium;
wherein the plurality of computer readable instructions, when executed by the processor, cause the processor to:
generating a first classified image matrix according to the first image matrix, wherein the first classified image matrix comprises a plurality of parts corresponding to a plurality of classes; and
obtaining a plurality of weights for first image processing corresponding to a plurality of parts of the first classified image matrix, and generating a first weight matrix according to the weights; and
an image processing unit, coupled to the receiving unit and the processor, for performing the first image processing on the first image matrix according to the first weight matrix to generate a first processed image matrix.
CN202011265432.7A 2020-11-13 2020-11-13 Image processing method and image processing system Pending CN114494691A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202011265432.7A CN114494691A (en) 2020-11-13 2020-11-13 Image processing method and image processing system

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202011265432.7A CN114494691A (en) 2020-11-13 2020-11-13 Image processing method and image processing system

Publications (1)

Publication Number Publication Date
CN114494691A true CN114494691A (en) 2022-05-13

Family

ID=81490279

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202011265432.7A Pending CN114494691A (en) 2020-11-13 2020-11-13 Image processing method and image processing system

Country Status (1)

Country Link
CN (1) CN114494691A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117079058A (en) * 2023-10-11 2023-11-17 腾讯科技（深圳）有限公司 Image processing method and device, storage medium and electronic equipment

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117079058A (en) * 2023-10-11 2023-11-17 腾讯科技（深圳）有限公司 Image processing method and device, storage medium and electronic equipment
CN117079058B (en) * 2023-10-11 2024-01-09 腾讯科技（深圳）有限公司 Image processing method and device, storage medium and electronic equipment

Similar Documents

Publication Publication Date Title
CN112288658B (en) Underwater image enhancement method based on multi-residual joint learning
CN110232661B (en) Low-illumination color image enhancement method based on Retinex and convolutional neural network
CN109934776B (en) Model generation method, video enhancement method, device and computer-readable storage medium
CN109427047B (en) Image processing method and device
US20070201756A1 (en) Image processing apparatus, mobile terminal device and image processing computer readable program
US11663707B2 (en) Method and system for image enhancement
CN110189260B (en) Image noise reduction method based on multi-scale parallel gated neural network
KR20170109898A (en) Apparatus and method for extracting object
CN112465727A (en) Low-illumination image enhancement method without normal illumination reference based on HSV color space and Retinex theory
US20130287299A1 (en) Image processing apparatus
CN114494691A (en) Image processing method and image processing system
WO2022226771A1 (en) Image processing method and image processing device
US11645739B2 (en) Image processing method and image processing system
US11082613B2 (en) Image adjusting method and image adjusting device
CN110545414B (en) Image sharpening method
KR102462265B1 (en) Directional scaling systems and methods
CN111383299B (en) Image processing method and device and computer readable storage medium
CN108304916B (en) Convolutional neural network optimization method combining attention mechanism and depth separable convolution
JP2000004360A (en) Reverse dither method of low complexity and low memory consumption
CN115619670A (en) Method, system and related equipment for enhancing low-light image
US10776076B2 (en) Neural network engine
KR101667306B1 (en) Low-Complexity Topological Derivative-Based Image Segmentation Method and System
KR102215607B1 (en) Electronic device capable of correction to improve the brightness of dark images and operating method thereof
US10198798B2 (en) Filtering device and filter method of the same
CN110060210A (en) Image processing method and relevant apparatus

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