CN110177277A - Image encoding method, device, computer readable storage medium and electronic equipment - Google Patents
Image encoding method, device, computer readable storage medium and electronic equipment Download PDFInfo
- Publication number
- CN110177277A CN110177277A CN201910580316.5A CN201910580316A CN110177277A CN 110177277 A CN110177277 A CN 110177277A CN 201910580316 A CN201910580316 A CN 201910580316A CN 110177277 A CN110177277 A CN 110177277A
- Authority
- CN
- China
- Prior art keywords
- image
- mapping table
- encoded
- global map
- unit
- 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.)
- Granted
Links
Classifications
-
- 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
- H04N19/124—Quantisation
-
- 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
- H04N19/134—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or criterion affecting or controlling the adaptive coding
- H04N19/167—Position within a video image, e.g. region of interest [ROI]
-
- 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
- H04N19/169—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding
- H04N19/182—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being a pixel
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Compression Or Coding Systems Of Tv Signals (AREA)
Abstract
The present invention provides a kind of image encoding method, device, computer readable storage medium and electronic equipments, are related to technical field of image processing.Image encoding method provided by the invention includes: to determine the corresponding image mapping table of image to be encoded based on image to be encoded；Determine multiple coding units of image to be encoded；The corresponding offset quantizing parameter sets of multiple coding units are determined based on image mapping table；Multiple coding units are performed the encoding operation based on offset quantizing parameter sets.The present invention solves in conventional images coding techniques, and there are obvious watershed areas between interesting image regions and regions of non-interest, and then leads to the problem of visual effect difference.
Description
Technical field
The present invention relates to technical field of image processing, and in particular to a kind of image encoding method, computer-readable is deposited device
Storage media and electronic equipment.
Background technique
In some image application scenarios or Video Applications scene including picture frame, some in image specifically feel emerging
Interesting region needs to be paid close attention to or emphasis processing, for example includes the region of facial information in image.
In the prior art, for the ease of paying close attention to or handling the area-of-interest of image, frame is selected usually in the form of rectangle frame
Area-of-interest, and improve the image quality of the area-of-interest of frame choosing.However, due to the image quality of area-of-interest and non-
There are marked differences for the image quality of area-of-interest, therefore, can deposit between the area-of-interest and regions of non-interest of image
In apparent watershed area, visual blocking artifact is in turn resulted in, user experience likability is very poor.
Summary of the invention
In view of this, the embodiment of the present invention is dedicated to providing a kind of image encoding method, device, computer-readable storage medium
Matter and electronic equipment, to solve in conventional images coding techniques, there are bright between interesting image regions and regions of non-interest
Aobvious watershed area, and then lead to the problem of visual effect difference.
In a first aspect, one embodiment of the invention provides a kind of image encoding method, the image encoding method include based on to
Coded image determines the corresponding image mapping table of image to be encoded；Determine multiple coding units of image to be encoded；Based on image
Mapping table determines the corresponding offset quantizing parameter sets of multiple coding units；It is single to multiple codings based on offset quantizing parameter sets
Member performs the encoding operation.
In an embodiment of the present invention, the corresponding image mapping table of image to be encoded is determined based on image to be encoded, including
Image to be encoded is input to mapping table model trained in advance, and determines the output result of mapping table model；It is tied based on output
Fruit determines the corresponding image mapping table of image to be encoded.
In an embodiment of the present invention, the corresponding offset quantizing parameter set of multiple coding units is determined based on image mapping table
It closes, the global map data including determining image mapping table；For coding unit, coding unit pair is determined based on image mapping table
The unit mapping data answered；Data are mapped based on global map data and unit, determine the corresponding offset of multiple coding units
Change parameter sets.
In an embodiment of the present invention, global map data include global map minimum value and global map average value, list
Member mapping data include unit mapping average value, map data based on global map data and unit, determine multiple coding units
Corresponding offset quantizing parameter sets, including it is directed to coding unit, determination unit maps average value and global map average value
Absolute difference；When absolute difference is greater than default mapping threshold value and unit mapping average value is less than global map average value,
The corresponding offset quantizing parameter of coding unit is equal to the quotient of the first difference and the second difference, wherein the first difference is unit mapping
The difference of average value and global map average value, the second difference are the difference of global map average value and global map minimum value；
The corresponding multiple offset quantizing parameter combinations of multiple coding units together, form offset quantizing parameter sets.
In an embodiment of the present invention, global map data include global map maximum value, based on global map data and
Unit maps data, determines the corresponding offset quantizing parameter sets of multiple coding units, further includes when absolute difference is greater than in advance
If map threshold value and unit mapping average value greater than global map average value, corresponding offset quantizing parameter of coding unit etc.
In the quotient of the first difference and third difference, wherein third difference is the difference of global map maximum value and global map average value.
In an embodiment of the present invention, mapping difference of the threshold value based on global map maximum value and global map minimum value is preset
Value determines.
In an embodiment of the present invention, the corresponding image mapping table of image to be encoded is determined based on image to be encoded, including
Based on the first interest information, the corresponding first image mapping table of image to be encoded is determined；Based on the second interest information, determine wait compile
The corresponding second image mapping table of code image；Mixing operation is carried out to the first image mapping table and the second image mapping table, with life
At the corresponding image mapping table of image to be encoded.
Second aspect, one embodiment of the invention provide a kind of picture coding device, which includes that image reflects
Firing table determining module, for determining the corresponding image mapping table of image to be encoded based on image to be encoded；Coding unit determines mould
Block, for determining multiple coding units of image to be encoded；Offset quantizing parameter sets determining module, for being mapped based on image
Table determines the corresponding offset quantizing parameter sets of multiple coding units；Coding module, for being based on offset quantizing parameter sets pair
Multiple coding units perform the encoding operation.
The third aspect, one embodiment of the invention provide a kind of computer readable storage medium, the computer-readable storage medium
Matter is stored with computer program, and the computer program is for executing image encoding method mentioned by any of the above-described embodiment.
Fourth aspect, one embodiment of the invention provide a kind of electronic equipment, which includes processor and for depositing
The memory of processor-executable instruction is stored up, the processor is for executing image coding staff mentioned by any of the above-described embodiment
Method.
Image encoding method provided in an embodiment of the present invention, by determine corresponding with image to be encoded image mapping table and
Multiple coding units are then based on image mapping table and determine the corresponding offset quantizing parameter sets of multiple coding units, and are based on
The mode that offset quantizing parameter sets perform the encoding operation multiple coding units realizes the coding behaviour for treating coded image
Make.Since image mapping table can characterize the degree of association of image and interest information to be encoded, what the embodiment of the present invention referred to
Image mapping table and coding unit are obtained into offset quantizing parameter sets to combining, and based on offset quantizing parameter sets to more
The mode that a coding unit performs the encoding operation can be realized the mesh that the interest information based on user adjusts the quality of coding unit
's.In addition, with existing rectangle circle select mode select area-of-interest method compared with, due to image mapping table include being capable of table
The degree of association data of more fine-grained image to be encoded and interest information are levied, therefore, the embodiment of the present invention can be realized image
In area-of-interest and regions of non-interest smoothly transit, and then improve visual effect, improve user experience good opinion
Degree.
Detailed description of the invention
Fig. 1 show the flow diagram of the image encoding method of one embodiment of the invention offer.
The image encoding method that Fig. 2 show one embodiment of the invention offer based on image to be encoded determines figure to be encoded
As the flow diagram of corresponding image mapping table step.
Fig. 3 show the practical application schematic diagram of the image encoding method of one embodiment of the invention offer.
The image encoding method that Fig. 4 show one embodiment of the invention offer based on image mapping table determines multiple codings
The flow diagram of the corresponding offset quantizing parameter sets step of unit.
Fig. 5 show mapping based on global map data and unit for the image encoding method of one embodiment of the invention offer
Data determine the flow diagram of the corresponding offset quantizing parameter sets step of multiple coding units.
Fig. 6 show another embodiment of the present invention provides image encoding method determined based on image to be encoded it is to be encoded
The flow diagram of the corresponding image mapping table step of image.
Fig. 7 show the structural schematic diagram of the picture coding device of one embodiment of the invention offer.
The structure that Fig. 8 show the image mapping table determining module of the picture coding device of one embodiment of the invention offer is shown
It is intended to.
Fig. 9 show the offset quantizing parameter sets determining module of the picture coding device of one embodiment of the invention offer
Structural schematic diagram.
Figure 10 show the offset quantizing parameter sets determination unit of the picture coding device of one embodiment of the invention offer
Structural schematic diagram.
Figure 11 show another embodiment of the present invention provides picture coding device image mapping table determining module knot
Structure schematic diagram.
Figure 12 show the structural schematic diagram of the electronic equipment of one embodiment of the invention offer.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation description, it is clear that the described embodiment is only a part of the embodiment of the present invention, instead of all the embodiments.Based on this
Embodiment in invention, every other reality obtained by those of ordinary skill in the art without making creative efforts
Example is applied, shall fall within the protection scope of the present invention.
Fig. 1 show the flow diagram of the image encoding method of one embodiment of the invention offer.As shown in Figure 1, this hair
The image encoding method that bright embodiment provides includes the following steps.
Step 10, the corresponding image mapping table of image to be encoded is determined based on image to be encoded.
Image mapping table refers to characterize the degree of association data acquisition system of the degree of association of image and interest information to be encoded,
The concrete form embodiment of the present invention of image mapping table is limited without unified.Wherein, interest information be user preset or
The information that person selectes, can sets itself according to the actual situation.For example, interest information is people's face facial information, then, figure to be encoded
Region as in including face facial information is area-of-interest, remaining region is regions of non-interest.
Step 20, multiple coding units of image to be encoded are determined.
In step 20, the data such as specific size and shape of coding unit can be set based on specific type of coding
It is fixed.For example, the size and shape of coding unit can be manually set in H.265 coding.For another example, in h .264, it encodes
H.264, the size and shape of unit, which is subject to, to be encoded.
Step 30, the corresponding offset quantizing parameter sets of multiple coding units are determined based on image mapping table.
It in step 30, include the corresponding multiple offset quantizing parameters of multiple coding units in offset quantizing parameter sets.
It should be appreciated that the corresponding encoder bit rate of coding unit can be adjusted based on offset quantizing parameter, and then adjust the product of coding unit
Matter.Wherein, the parameter that quality can characterize picture quality based on clarity etc. is measured.
Step 40, multiple coding units are performed the encoding operation based on offset quantizing parameter sets.
The encoding operation referred in step 40 both can be H.265 encoding operation, and can also be H.264 encoding operation, this
Inventive embodiments limit this without unified.
In actual application, it is primarily based on image to be encoded and determines the corresponding image mapping table of image to be encoded, so
Multiple coding units of image to be encoded are determined afterwards, and the corresponding offset quantizing of multiple coding units is determined based on image mapping table
Parameter sets finally perform the encoding operation multiple coding units based on offset quantizing parameter sets.
Image encoding method provided in an embodiment of the present invention, by determine corresponding with image to be encoded image mapping table and
Multiple coding units are then based on image mapping table and determine the corresponding offset quantizing parameter sets of multiple coding units, and are based on
The mode that offset quantizing parameter sets perform the encoding operation multiple coding units realizes the coding behaviour for treating coded image
Make.Since image mapping table can characterize the degree of association of image and interest information to be encoded, what the embodiment of the present invention referred to
Image mapping table and coding unit are obtained into offset quantizing parameter sets to combining, and based on offset quantizing parameter sets to more
The mode that a coding unit performs the encoding operation can be realized the mesh that the interest information based on user adjusts the quality of coding unit
's.In addition, with existing rectangle circle select mode select area-of-interest method compared with, due to image mapping table include being capable of table
The degree of association data of more fine-grained image to be encoded and interest information are levied, therefore, the embodiment of the present invention can be realized image
In area-of-interest and regions of non-interest smoothly transit, and then improve visual effect, improve user experience good opinion
Degree.
The image encoding method that Fig. 2 show one embodiment of the invention offer based on image to be encoded determines figure to be encoded
As the flow diagram of corresponding image mapping table step.Extend shown in Fig. 2 on the basis of embodiment illustrated in fig. 1 of the present invention
Embodiment, describes emphatically the difference of embodiment illustrated in fig. 2 and embodiment illustrated in fig. 1 below, and something in common repeats no more.
As shown in Fig. 2, in embodiments of the present invention, determining that the corresponding image of image to be encoded maps based on image to be encoded
(i.e. step 10) includes the following steps table step.
Step 11, image to be encoded is input to mapping table model trained in advance, and determines the output of mapping table model
As a result.
It should be noted that the mapping table model referred in step 11, can be based on area-of-interest sample training and
At machine learning map table model.
Step 12, the corresponding image mapping table of image to be encoded is determined based on output result.
In actual application, image to be encoded is input to mapping table model trained in advance first, and determination is reflected
The output of firing table model as a result, and the corresponding image mapping table of image to be encoded is determined based on output result, then determine wait compile
Multiple coding units of code image, and the corresponding offset quantizing parameter sets of multiple coding units are determined based on image mapping table,
Finally multiple coding units are performed the encoding operation based on offset quantizing parameter sets.
Image encoding method provided in an embodiment of the present invention determines image to be encoded using mapping table model trained in advance
Corresponding image mapping table, it is determined that the image mapping table with the interest information matched of user, and then precisely to distinguish figure
As area-of-interest and regions of non-interest provide precondition.
Table 1 show the practical application schematic diagram of the mapping table model of one embodiment of the invention offer.Specifically, at this
In inventive embodiments, mapping table model is based on convolutional neural networks structure mould made of multiple semi-cylindrical hills sample training
Type, the size of image to be encoded are 224*224*3 through bilinear interpolation scaling size.As shown in table 1, table 1 shows the present invention
The main hierarchical structure for the mapping table model that embodiment provides, in embodiments of the present invention, mapping table model mainly includes convolution
Layer and pond layer (i.e. max-pool) etc..
Table 1 maps table model
| Main thoroughfare | Image (224*224*3) to be encoded |
| 3*3*64, s=1 | |
| 3*3*64, s=1 | |
| Max-pool, s=2 | |
| 3*3*128, s=1 | |
| 3*3*128, s=1 | |
| Max-pool, s=2 | |
| 3*3*256, s=1 | |
| 3*3*256, s=1 | |
| 3*3*256, s=1 | |
| Max-pool, s=2 | |
| 3*3*512, s=1 | |
| 3*3*512, s=1 | |
| 3*3*512, s=1 | |
| Max-pool, s=2 | |
| 3*3*512, s=1 | |
| 3*3*512, s=1 | |
| 3*3*512, s=1 | [14][14][512] |
| 3*3*32, s=1 | |
| 3*3*1024, s=1 | |
| 3*3*1024, s=1 | Last convolution |
| reduce-mean[1,2] | gap[1024] |
| gap*W[1024][257] | class-prob[257] |
In actual application, the image to be encoded of 224*224*3 is input in mapping table model, treats volume first
Code image carry out twice output channel number be 64, the convolution operation of convolution kernel 3*3, step-length s=1, and carry out a step-length s=
2 maximum value ponds (i.e. max-pool) are then and by output channel number twice 128, convolution kernel 3*3, step-length s=1
Convolution operation, and a step-length s=2 maximum value pond is carried out, then using output channel number is 256, convolution kernel is three times
The convolution operation of 3*3, step-length s=1, and a step-length s=2 maximum value pond is carried out, then using output channel number three times
For the convolution operation of 512, convolution kernel 3*3, step-length s=1, and carry out a step-length s=2 maximum value pond, then using
Three times output channel number be 512, the convolution operation of convolution kernel 3*3, step-length s=1, to obtain the characteristic pattern of 14*14*512
As (i.e. port number is 512, and the picture size in each channel is 14*14), then using cubic convolution, (i.e. port number is respectively
32,1024,1024, convolution kernel is 3*3, step-length s=1), to obtain the characteristic image of 14*14*1024, (i.e. port number is
1024, the picture size in each channel is 14*14).
Based on the characteristic image of obtained 14*14*1024, based on being averaged for each channel of corresponding calculating reduce-mean
Value, to obtain gap [1024] a value, finally to gap [1024] multiplied by weight coefficient (i.e. W [1024] [257]), and then obtains
The output probability of 257 classes.In embodiments of the present invention, 257 class probabilities respectively correspond 256 classifications and 1 does not include
The case where classification.
So, in the actual application of one embodiment of the invention, the output probability of 257 classes is ranked up, and
Determine that multiple classes of maximum probability, such as the embodiment of the present invention use 5 maximum probability classes, and 5 based on maximum probability classes
Fusion generates the image mapping table of the area-of-interest of image to be encoded.For example, the class of maximum probability is 50 classes, then, first
The weight coefficient w [1024] that the 50th class is extracted from the weight coefficient w [1024] [257] of all classes, then utilize with it is to be encoded
The last one convolutional layer [14] [14] [1024] is adjusted to [224] [224] [1024] by the identical bilinear interpolation of image size
(along axis [1,2]) is then readjusted [224] [224] [1024] for [50176] [1024], finally by [50176] [1024]
Matrix is multiplied with weight coefficient vector w [1024], obtains the area-of-interest exposure mask diagram data of [224] [244], and normalization calculates
The mapping data of an area-of-interest are generated afterwards, and the corresponding mapping data of 5 classes are then carried out fusion and generate the present invention in fact
Apply the corresponding image mapping table of image to be encoded that example refers to.
It should be noted that the mapping table model that 1 illustrated embodiment of table refers to is merely illustrative, not to limit to
The present invention.Other convolutional neural networks structural models or other calculations can also be applied to calculate in the embodiment of the present invention
The image mapping table referred to, the present invention are no longer described in detail.
Fig. 3 show the practical application schematic diagram of the image encoding method of one embodiment of the invention offer.Such as
Shown in Fig. 3, in embodiments of the present invention, the image encoding method that the above embodiment of the present invention is referred to is applied to figure
As in encoding operation, i.e., generating image corresponding with image to be encoded using image mapping table generation module shown in Fig. 3 and mapping
Table, and image mapping table generated is applied in the encoding operation of image to be encoded.Specifically, in the embodiment of the present invention
Image to be encoded is rgb format image.
In actual application, the image to be encoded of rgb format is converted into YUV420 format first, be then based on to
Coded image generates image mapping table corresponding with image to be encoded, and combines image mapping table to the to be encoded of YUV420 format
The operations such as image quantified, inverse quantization, intra-prediction, the image after ultimately producing coding.Wherein, image in encoded
In, the quality of area-of-interest is elevated and/or the quality of regions of non-interest is lowered, also, area-of-interest and non-sense
It realizes and smoothly transits between interest region.
It should be noted that the image coding scene that the image mapping table that embodiment illustrated in fig. 3 refers to is applied to is only
Exemplary illustration, not to limit to the present invention.The figure that other images coding scene can also be referred to using the embodiment of the present invention
As mapping table, the present invention is no longer described in detail.
The image encoding method that Fig. 4 show one embodiment of the invention offer based on image mapping table determines multiple codings
The flow diagram of the corresponding offset quantizing parameter sets step of unit.Extend on the basis of embodiment illustrated in fig. 1 of the present invention
Embodiment illustrated in fig. 4 out describes emphatically the difference of embodiment illustrated in fig. 4 and embodiment illustrated in fig. 1 below, and something in common is not
It repeats again.
As shown in figure 4, in embodiments of the present invention, determining the corresponding offset of multiple coding units based on image mapping table
Changing parameter sets step, (i.e. step 30) includes the following steps.
Step 31, the global map data of image mapping table are determined.
Global map data refer to characterize the data of the Global Information of image to be encoded.For example, in the present invention one
In embodiment, global map data include global map minimum value, global map average value and global map maximum value, wherein
Global map minimum value refers to the minimum value for all data for including in image mapping table, and global map average value refers to figure
As the average value for all data for including in mapping table, global map maximum value refers to all numbers for including in image mapping table
According to maximum value.
Step 32, for coding unit, determine that the corresponding unit of coding unit maps data based on image mapping table.
Unit mapping data refer to characterize the data of the encoded element information of image to be encoded.For example, in this hair
In a bright embodiment, unit mapping data include unit mapping average value, wherein unit mapping average value refers to coding unit
The average value of all data in corresponding image mapping table.
Step 33, data are mapped based on global map data and unit, determines the corresponding offset quantizing of multiple coding units
Parameter sets.
In actual application, it is primarily based on image to be encoded and determines the corresponding image mapping table of image to be encoded, so
Multiple coding units of image to be encoded are determined afterwards, and the corresponding offset quantizing of multiple coding units is determined based on image mapping table
Then parameter sets determine the global map data of image mapping table, for coding unit, determined and encoded based on image mapping table
The corresponding unit of unit maps data, finally maps data based on global map data and unit, determines multiple coding units pair
The offset quantizing parameter sets answered.
Image encoding method provided in an embodiment of the present invention, by determining that global map data and unit map data, so
The mode of the corresponding offset quantizing parameter sets of multiple coding units is determined based on global map data and unit mapping data afterwards,
Realize the purpose that the corresponding offset quantizing parameter sets of multiple coding units are determined based on image mapping table.Due to global map
Data can characterize the global feature of image mapping table, and unit mapping data can characterize the corresponding image mapping table of coding unit
The local feature in region, therefore, the embodiment of the present invention can more accurately be determined based on the actual conditions of image to be encoded and be encoded
The corresponding offset quantizing parameter sets of unit, and then to further realize area-of-interest in image to be encoded and non-interested
Smoothly transitting for region provides precondition, to further improve visual effect.
Fig. 5 show mapping based on global map data and unit for the image encoding method of one embodiment of the invention offer
Data determine the flow diagram of the corresponding offset quantizing parameter sets step of multiple coding units.It is real shown in Fig. 4 of the present invention
It applies and extends embodiment illustrated in fig. 5 on the basis of example, narration embodiment illustrated in fig. 5 and embodiment illustrated in fig. 4 be not emphatically below
Same place, something in common repeat no more.
As shown in figure 5, in embodiments of the present invention, mapping data based on global map data and unit, determining multiple volumes
(i.e. step 33) includes the following steps the corresponding offset quantizing parameter sets step of code unit.
Step 331, for coding unit, determination unit maps the absolute difference of average value and global map average value.
Step 332, judge whether absolute difference is greater than default mapping threshold value.
It should be noted that default mapping threshold value specifically can sets itself according to the actual situation, the embodiment of the present invention is to this
It is limited without unified.
For example, in an embodiment of the present invention, presetting mapping threshold value and being based on global map maximum value and global map minimum
The difference of value determines.Since global map maximum value and global map minimum value can embody the extremum characteristic of image mapping table,
Therefore, the embodiment of the present invention can accurately determine the corresponding offset quantizing parameter sets of coding unit for subsequent fine and provide premise item
Part effectively prevents the generation of over-deflection and excessive the case where adjusting.
In step 332, when judging that absolute difference is less than or equal to default mapping threshold value, 334 are thened follow the steps, when
When judging that absolute difference is greater than default mapping threshold value, 333 are thened follow the steps.
Step 333, judge whether the unit mapping average value of coding unit is less than global map average value.
In step 333, when the unit for judging coding unit, which maps average value, is less than global map average value, then execute
Step 335, it when the unit for judging coding unit, which maps average value, is greater than or equal to global map average value, thens follow the steps
336。
Step 334, determine that the corresponding offset quantizing parameter of coding unit is equal to zero.
Step 335, determine that the corresponding offset quantizing parameter of coding unit is equal to the quotient of the first difference and the second difference.
In step 335, the first difference is the difference that unit maps average value and global map average value, and the second difference is
The difference of global map average value and global map minimum value.
Step 336, determine that the corresponding offset quantizing parameter of coding unit is equal to the quotient of the first difference and third difference.
In step 336, third difference is the difference of global map maximum value and global map average value.
It should be noted that in embodiments of the present invention, when offset quantizing parameter is equal to zero, then maintaining coding unit pair
The image-region quality answered is constant, when offset quantizing parameter is less than zero, then reduces the corresponding image-region quality of coding unit,
The quality for reducing regions of non-interest then improves the corresponding image-region of coding unit when offset quantizing parameter is greater than zero
Quality improves the quality of area-of-interest.
In actual application, it is primarily based on image to be encoded and determines the corresponding image mapping table of image to be encoded, so
Multiple coding units of image to be encoded are determined afterwards, and the corresponding offset quantizing of multiple coding units is determined based on image mapping table
Then parameter sets determine the global map data of image mapping table, for coding unit, determined and encoded based on image mapping table
The corresponding unit of unit maps data, then for the difference of coding unit determination unit mapping average value and global map average value
It is worth absolute value, and judges whether absolute difference is greater than default mapping threshold value, is preset when judging that absolute difference is less than or equal to
It when mapping threshold value, determines that the corresponding offset quantizing parameter of coding unit is equal to zero, presets mapping when judging that absolute difference is greater than
When threshold value, judge whether the unit mapping average value of coding unit is less than global map average value, when the list for judging coding unit
Member mapping average value be less than global map average value when, determine the corresponding offset quantizing parameter of coding unit be equal to the first difference and
The quotient of second difference determines and compiles when the unit for judging coding unit, which maps average value, is greater than or equal to global map average value
The corresponding offset quantizing parameter of code unit is equal to the quotient of the first difference and third difference.
Image encoding method provided in an embodiment of the present invention, by judging whether absolute difference is greater than default mapping threshold
Value is then judged whether the unit mapping average value of coding unit is less than global map average value, and is obtained based on judging result
The mode of the method for determination of different offset quantizing parameters realizes more based on global map data and unit mapping data determination
The purpose of the corresponding offset quantizing parameter sets of a coding unit, and then realize the reality based on image to be encoded and coding unit
Border situation precisely determines the purpose of the corresponding offset quantizing parameter sets of coding unit.Therefore, the embodiment of the present invention is final real
Area-of-interest and smoothly transitting for regions of non-interest in existing image provide precondition.In addition, the embodiment of the present invention is logical
It crosses and judges whether the unit mapping average value of coding unit is less than global map average value, and work as and judge that the unit of coding unit reflects
When penetrating average value less than global map average value, determine that the corresponding offset quantizing parameter of coding unit is equal to the first difference and second
The mode of the quotient of difference realizes the quality that regions of non-interest is reduced in the case where maintaining area-of-interest quality constant
Purpose, has not only highlighted the visual effect of area-of-interest as a result, and reduces the encoder bit rate of image to be encoded.
Fig. 6 show another embodiment of the present invention provides image encoding method determined based on image to be encoded it is to be encoded
The flow diagram of the corresponding image mapping table step of image.Extend Fig. 6 institute on the basis of embodiment illustrated in fig. 1 of the present invention
Show embodiment, describe the difference of embodiment illustrated in fig. 6 and embodiment illustrated in fig. 1 emphatically below, something in common repeats no more.
As shown in fig. 6, in embodiments of the present invention, determining that the corresponding image of image to be encoded maps based on image to be encoded
(i.e. step 10) includes the following steps table step.
Step 13, it is based on the first interest information, determines the corresponding first image mapping table of image to be encoded.
Step 14, it is based on the second interest information, determines the corresponding second image mapping table of image to be encoded.
Step 15, mixing operation is carried out to the first image mapping table and the second image mapping table, to generate image to be encoded
Corresponding image mapping table.
The second interest information referred in the first interest information and step 14 referred in step 13 is different types of emerging
Interesting information, the first interest information and the second interest information can be according to image to be encoded and practical application scene sets itselfs.
It, can also sets itself amalgamation mode and fusion parameters according to the actual situation in addition, the mixing operation referred in step 15.
In actual application, it is primarily based on the first interest information and determines the corresponding first image mapping of image to be encoded
Table, and the corresponding second image mapping table of image to be encoded is determined based on the second interest information, then to the first image mapping table
Mixing operation, which is carried out, with the second image mapping table determines image to be encoded to generate the corresponding image mapping table of image to be encoded
Multiple coding units, and the corresponding offset quantizing parameter sets of multiple coding units, last base are determined based on image mapping table
Multiple coding units are performed the encoding operation in offset quantizing parameter sets.
Image encoding method provided in an embodiment of the present invention, by determining that the first image maps based on the first interest information
Table, and the second image mapping table is determined based on the second interest information, and then map based on the first image mapping table and the second image
Table realizes the area-of-interest based on polymorphic type interest information in a manner of generating the corresponding image mapping table of image to be encoded
With smoothly transitting for regions of non-interest, to further improve visual effect, and user experience good opinion is further improved
Degree.
It should be appreciated that the number of types of interest information, is not limited to the above embodiment two referred to, or three
It is a, four or more.
In an embodiment of the present invention, the number of types of interest information is multiple, also, during mixing operation, into
Mixing operation of the row based on weight.Wherein, the specific set-up mode of weight and specific setting parameter can according to the actual situation certainly
Row setting.
Fig. 7 show the structural schematic diagram of the picture coding device of one embodiment of the invention offer.As shown in fig. 7, this hair
The picture coding device that bright embodiment provides includes:
Image mapping table determining module 100, for determining that the corresponding image of image to be encoded maps based on image to be encoded
Table；
Coding unit determining module 200, for determining multiple coding units of image to be encoded；
Offset quantizing parameter sets determining module 300, for determining that multiple coding units are corresponding based on image mapping table
Offset quantizing parameter sets；
Coding module 400, for being performed the encoding operation based on offset quantizing parameter sets to multiple coding units.
The structure that Fig. 8 show the image mapping table determining module of the picture coding device of one embodiment of the invention offer is shown
It is intended to.Extend embodiment illustrated in fig. 8 on the basis of embodiment illustrated in fig. 7 of the present invention, implements shown in narration Fig. 8 emphatically below
The difference of example and embodiment illustrated in fig. 7, something in common repeat no more.
As shown in figure 8, image mapping table determining module 100 is wrapped in picture coding device provided in an embodiment of the present invention
It includes:
Result determination unit 110 is exported, for image to be encoded to be input to mapping table model trained in advance, and is determined
Map the output result of table model；
Image mapping table determination unit 120, for determining the corresponding image mapping table of image to be encoded based on output result.
Fig. 9 show the offset quantizing parameter sets determining module of the picture coding device of one embodiment of the invention offer
Structural schematic diagram.Extend embodiment illustrated in fig. 9 on the basis of embodiment illustrated in fig. 7 of the present invention, describes Fig. 9 institute emphatically below
Show that the difference of embodiment and embodiment illustrated in fig. 7, something in common repeat no more.
As shown in figure 9, in picture coding device provided in an embodiment of the present invention, offset quantizing parameter sets determining module
300 include:
Global map data determination unit 310, for determining the global map data of image mapping table；
Unit maps data determination unit 320, for being directed to coding unit, determines coding unit pair based on image mapping table
The unit mapping data answered；
Offset quantizing parameter sets determination unit 330, for mapping data based on global map data and unit, determination is more
The corresponding offset quantizing parameter sets of a coding unit.
Figure 10 show the offset quantizing parameter sets determination unit of the picture coding device of one embodiment of the invention offer
Structural schematic diagram.Extend embodiment illustrated in fig. 10 on the basis of embodiment illustrated in fig. 9 of the present invention, narration is schemed emphatically below
The difference of 10 illustrated embodiments and embodiment illustrated in fig. 9, something in common repeat no more.
As shown in Figure 10, in picture coding device provided in an embodiment of the present invention, offset quantizing parameter sets determine single
First 330 include:
Absolute difference determines subelement 3310, and for being directed to coding unit, determination unit mapping average value is reflected with the overall situation
Penetrate the absolute difference of average value；
First judgment sub-unit 3320, for judging whether absolute difference is greater than default mapping threshold value；
Second judgment sub-unit 3330, for judging that it is whether flat less than global map that the unit of coding unit maps average value
Mean value；
First offset quantizing parameter determines subelement 3340, for determining that the corresponding offset quantizing parameter of coding unit is equal to
Zero；
Second offset quantizing parameter determines subelement 3350, for determining that the corresponding offset quantizing parameter of coding unit is equal to
The quotient of first difference and the second difference；
Third offset quantizing parameter determines subelement 3360, for determining that the corresponding offset quantizing parameter of coding unit is equal to
The quotient of first difference and third difference.
Figure 11 show another embodiment of the present invention provides picture coding device image mapping table determining module knot
Structure schematic diagram.Extend embodiment illustrated in fig. 11 on the basis of embodiment illustrated in fig. 7 of the present invention, describes Figure 11 institute emphatically below
Show that the difference of embodiment and embodiment illustrated in fig. 7, something in common repeat no more.
As shown in figure 11, in picture coding device provided in an embodiment of the present invention, image mapping table determining module 100 is wrapped
It includes:
First image mapping table determination unit 130 determines image to be encoded corresponding for being based on the first interest information
One image mapping table；
Second image mapping table determination unit 140 determines image to be encoded corresponding for being based on the second interest information
Two image mapping tables；
Integrated unit 150, for carrying out mixing operation to the first image mapping table and the second image mapping table, with generate to
The corresponding image mapping table of coded image.
It should be noted that image mapping table determining module 100, volume in the picture coding device that Fig. 7 to Figure 11 is provided
Code unit determining module 200, offset quantizing parameter sets determining module 300 and coding module 400 and image mapping table determine
Output result determination unit 110, image mapping table determination unit 120, the first image mapping table determination unit in module 100
130, in the second image mapping table determination unit 140 and integrated unit 150 and offset quantizing parameter sets determining module 300
Global map data determination unit 310, unit mapping data determination unit 320 and offset quantizing parameter sets determination unit
330 and offset quantizing parameter sets determination unit 330 in absolute difference determine that judgement of subelement 3310, first is single
First 3320, second judgment sub-unit 3330, the first offset quantizing parameter determine that subelement 3340, the second offset quantizing parameter determine
Subelement 3350 and third offset quantizing parameter determine that the operations and functions of subelement 3360 can be mentioned with reference to above-mentioned Fig. 1 to Fig. 6
The image encoding method of confession, in order to avoid repeating, details are not described herein.
In the following, being described with reference to Figure 12 the electronic equipment for realizing image encoding method provided in an embodiment of the present invention.Figure 12
The structural schematic diagram of the electronic equipment of one embodiment of the invention offer is provided.As shown in figure 12, electronic equipment 50 includes one
Or multiple processors 501 and memory 502.
Processor 501 can be central processing unit (CPU) or have data-handling capacity and/or instruction execution capability
Other forms processing unit, and can control the other assemblies in electronic equipment to execute desired function.
Memory 502 may include one or more computer program products, and the computer program product may include
Various forms of computer readable storage mediums, such as volatile memory and/or nonvolatile memory.The volatibility is deposited
Reservoir for example may include random access memory (RAM) and/or cache memory (cache) etc..It is described non-volatile
Memory for example may include read-only memory (ROM), hard disk, flash memory etc..It can be on the computer readable storage medium
One or more computer program instructions are stored, processor 501 can run described program instruction, to realize the above-mentioned reality of the present invention
Apply the image encoding method and/or other desired functions that example refers to.May be used also in the computer readable storage medium
With various contents such as storage band coded images.
In one example, electronic equipment 50 can also include input unit 503 and output device 504.These components are logical
Cross bindiny mechanism's (not shown) interconnection of bus system and/or other forms.Wherein, input unit 503 includes but unlimited
In for keyboard, mouse etc..
Certainly, it to put it more simply, illustrating only some in component related to the present invention in the electronic equipment in Figure 12, saves
The component of such as bus, input/output interface etc. is omited.In addition to this, according to concrete application situation, electronic equipment can be with
Including any other component appropriate.
Other than the above method and equipment, the embodiment of the present invention can also be computer program product comprising meter
Calculation machine program instruction, it is above-mentioned that the computer program instructions make the processor execute this specification when being run by processor
The step in image encoding method that the embodiment of the present invention described in " illustrative methods " part refers to.
The computer program product can be write with any combination of one or more programming languages for holding
The program code of row embodiment of the present invention operation, described program design language includes object oriented program language, such as
Java, C++ etc. further include conventional procedural programming language, such as " C " language or similar programming language.Journey
Sequence code can be executed fully on the user computing device, partly execute on a user device, be independent soft as one
Part packet executes, part executes on a remote computing or completely in remote computing device on the user computing device for part
Or it is executed on server.
In addition, the embodiment of the present invention can also be computer readable storage medium, computer program instructions are stored thereon with,
It is above-mentioned " illustrative methods " that the computer program instructions make the processor execute this specification when being run by processor
Step in image encoding method described in part.
The computer readable storage medium can be using any combination of one or more readable mediums.Readable medium can
To be readable signal medium or readable storage medium storing program for executing.Readable storage medium storing program for executing for example can include but is not limited to electricity, magnetic, light, electricity
Magnetic, the system of infrared ray or semiconductor, device or device, or any above combination.Readable storage medium storing program for executing it is more specific
Example (non exhaustive list) includes: the electrical connection with one or more conducting wires, portable disc, hard disk, random access memory
Device (RAM), read-only memory (ROM), erasable programmable read only memory (EPROM or flash memory), optical fiber, portable compact disc
Read-only memory (CD-ROM), light storage device, magnetic memory device or above-mentioned any appropriate combination.
Basic principle of the invention is described in conjunction with specific embodiments above, however, it is desirable to, it is noted that in the present invention
The advantages of referring to, advantage, effect etc. are only exemplary rather than limitation, must not believe that these advantages, advantage, effect etc. are of the invention
Each embodiment is prerequisite.In addition, detail disclosed above is merely to exemplary effect and the work being easy to understand
With, rather than limit, above-mentioned details is not intended to limit the present invention as that must realize using above-mentioned concrete details.
Device involved in the present invention, device, equipment, system block diagram only as illustrative example and be not intended to
It is required that or hint must be attached in such a way that box illustrates, arrange, configure.As those skilled in the art will appreciate that
, it can be connected by any way, arrange, configure these devices, device, equipment, system.Such as "include", "comprise", " tool
" etc. word be open vocabulary, refer to " including but not limited to ", and can be used interchangeably with it.Vocabulary used herein above
"or" and "and" refer to vocabulary "and/or", and can be used interchangeably with it, unless it is not such that context, which is explicitly indicated,.Here made
Vocabulary " such as " refers to phrase " such as, but not limited to ", and can be used interchangeably with it.
It may also be noted that each component or each step are can to decompose in the device of the invention, device and method
And/or reconfigure.These, which decompose and/or reconfigure, should be regarded as equivalent scheme of the invention.
The foregoing is merely illustrative of the preferred embodiments of the present invention, is not intended to limit the invention, all in essence of the invention
Within mind and principle, made any modification, equivalent replacement etc. be should all be included in the protection scope of the present invention.
Claims (10)
1. a kind of image encoding method characterized by comprising
The corresponding image mapping table of the image to be encoded is determined based on image to be encoded；
Determine multiple coding units of the image to be encoded；
The corresponding offset quantizing parameter sets of the multiple coding unit are determined based on described image mapping table；
The multiple coding unit is performed the encoding operation based on the offset quantizing parameter sets.
2. image encoding method according to claim 1, which is characterized in that it is described based on image to be encoded determine it is described to
The corresponding image mapping table of coded image, comprising:
The image to be encoded is input to mapping table model trained in advance, and determines the output knot of the mapping table model
Fruit；
The corresponding image mapping table of the image to be encoded is determined based on the output result.
3. image encoding method according to claim 1 or 2, which is characterized in that described true based on described image mapping table
Determine the corresponding offset quantizing parameter sets of the multiple coding unit, comprising:
Determine the global map data of described image mapping table；
For the coding unit, determine that the corresponding unit of the coding unit maps data based on described image mapping table；
Data are mapped based on the global map data and the unit, determine the corresponding offset quantizing of the multiple coding unit
Parameter sets.
4. image encoding method according to claim 3, which is characterized in that the global map data include global map
Minimum value and global map average value, the unit mapping data include unit mapping average value, described to be reflected based on the overall situation
Data and unit mapping data are penetrated, determine the corresponding offset quantizing parameter sets of the multiple coding unit, comprising:
For the coding unit, the absolute difference of unit the mapping average value and the global map average value is determined；
When the absolute difference is greater than default mapping threshold value and the unit maps average value and is averaged less than the global map
When value, the corresponding offset quantizing parameter of the coding unit is equal to the quotient of the first difference and the second difference, wherein described the
One difference is the difference that the unit maps average value and the global map average value, and second difference is that the overall situation is reflected
Penetrate the difference of average value and the global map minimum value；
The corresponding the multiple offset quantizing parameter combination of the multiple coding unit together, forms the offset quantizing ginseng
Manifold is closed.
5. image encoding method according to claim 4, which is characterized in that the global map data include global map
Maximum value, it is described to map data based on the global map data and the unit, determine that the multiple coding unit is corresponding
Offset quantizing parameter sets, further includes:
When the absolute difference is greater than default mapping threshold value and the unit maps average value and is averaged greater than the global map
When value, the corresponding offset quantizing parameter of the coding unit is equal to the quotient of the first difference and third difference, wherein described the
Three differences are the difference of the global map maximum value and the global map average value.
6. image encoding method according to claim 4 or 5, which is characterized in that the default mapping threshold value is based on described
The difference of global map maximum value and the global map minimum value determines.
7. image encoding method according to claim 1 or 2, which is characterized in that described to determine institute based on image to be encoded
State the corresponding image mapping table of image to be encoded, comprising:
Based on the first interest information, the corresponding first image mapping table of the image to be encoded is determined；
Based on the second interest information, the corresponding second image mapping table of the image to be encoded is determined；
Mixing operation is carried out to the first image mapping table and the second image mapping table, it is corresponding to generate image to be encoded
Image mapping table.
8. a kind of picture coding device characterized by comprising
Image mapping table determining module, for determining the corresponding image mapping table of the image to be encoded based on image to be encoded；
Coding unit determining module, for determining multiple coding units of the image to be encoded；
Offset quantizing parameter sets determining module, for determining that the multiple coding unit is corresponding based on described image mapping table
Offset quantizing parameter sets；
Coding module, for being performed the encoding operation based on the offset quantizing parameter sets to the multiple coding unit.
9. a kind of computer readable storage medium, the computer-readable recording medium storage has computer program, the calculating
Machine program is used to execute any image encoding method of the claims 1-7.
10. a kind of electronic equipment, the electronic equipment include:
Processor；
For storing the memory of the processor-executable instruction；
The processor, for executing any image encoding method of the claims 1-7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910580316.5A CN110177277B (en) | 2019-06-28 | 2019-06-28 | Image coding method and device, computer readable storage medium and electronic equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910580316.5A CN110177277B (en) | 2019-06-28 | 2019-06-28 | Image coding method and device, computer readable storage medium and electronic equipment |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110177277A true CN110177277A (en) | 2019-08-27 |
| CN110177277B CN110177277B (en) | 2022-04-12 |
Family
ID=67699472
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910580316.5A Active CN110177277B (en) | 2019-06-28 | 2019-06-28 | Image coding method and device, computer readable storage medium and electronic equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110177277B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110769254A (en) * | 2019-10-10 | 2020-02-07 | 网宿科技股份有限公司 | Code rate configuration method, system and equipment for video frame |
| WO2021098030A1 (en) * | 2019-11-22 | 2021-05-27 | 网宿科技股份有限公司 | Method and apparatus for video encoding |
| CN113824967A (en) * | 2021-11-23 | 2021-12-21 | 环球数科集团有限公司 | Video compression method based on deep learning |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2016054307A1 (en) * | 2014-10-03 | 2016-04-07 | Microsoft Technology Licensing, Llc | Adapting quantization within regions-of-interest |
| CN106507110A (en) * | 2016-12-06 | 2017-03-15 | 北京华夏电通科技有限公司 | Court's trial witness picture and video motion picture guard method, device and system |
| CN106713915A (en) * | 2015-11-16 | 2017-05-24 | 三星电子株式会社 | Method of encoding video data |
| CN107534768A (en) * | 2015-03-02 | 2018-01-02 | 三星电子株式会社 | Method and apparatus for being compressed based on photographing information to image |
| US20180376153A1 (en) * | 2017-06-21 | 2018-12-27 | Intel Corporation | Content, psychovisual, region of interest, and persistence based adaptive quantization for video coding |
| CN109429066A (en) * | 2017-08-29 | 2019-03-05 | 三星电子株式会社 | Video coding apparatus and video coding system |
| CN109889839A (en) * | 2019-03-27 | 2019-06-14 | 上海交通大学 | ROI Image Coding, decoding system and method based on deep learning |
-
2019
- 2019-06-28 CN CN201910580316.5A patent/CN110177277B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2016054307A1 (en) * | 2014-10-03 | 2016-04-07 | Microsoft Technology Licensing, Llc | Adapting quantization within regions-of-interest |
| CN107534768A (en) * | 2015-03-02 | 2018-01-02 | 三星电子株式会社 | Method and apparatus for being compressed based on photographing information to image |
| CN106713915A (en) * | 2015-11-16 | 2017-05-24 | 三星电子株式会社 | Method of encoding video data |
| CN106507110A (en) * | 2016-12-06 | 2017-03-15 | 北京华夏电通科技有限公司 | Court's trial witness picture and video motion picture guard method, device and system |
| US20180376153A1 (en) * | 2017-06-21 | 2018-12-27 | Intel Corporation | Content, psychovisual, region of interest, and persistence based adaptive quantization for video coding |
| CN109429066A (en) * | 2017-08-29 | 2019-03-05 | 三星电子株式会社 | Video coding apparatus and video coding system |
| CN109889839A (en) * | 2019-03-27 | 2019-06-14 | 上海交通大学 | ROI Image Coding, decoding system and method based on deep learning |
Non-Patent Citations (1)
| Title |
|---|
| 王明慧: "基于H.264的感兴趣区域视频编码研究", 《中国优秀硕士论文数据库》 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110769254A (en) * | 2019-10-10 | 2020-02-07 | 网宿科技股份有限公司 | Code rate configuration method, system and equipment for video frame |
| CN110769254B (en) * | 2019-10-10 | 2022-04-22 | 网宿科技股份有限公司 | Code rate configuration method, system and equipment for video frame |
| WO2021098030A1 (en) * | 2019-11-22 | 2021-05-27 | 网宿科技股份有限公司 | Method and apparatus for video encoding |
| CN113824967A (en) * | 2021-11-23 | 2021-12-21 | 环球数科集团有限公司 | Video compression method based on deep learning |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110177277B (en) | 2022-04-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110177277A (en) | Image encoding method, device, computer readable storage medium and electronic equipment | |
| US11915133B2 (en) | Techniques for smooth region merging in image editing | |
| CN109816011A (en) | Generate the method and video key frame extracting method of portrait parted pattern | |
| CN113822437B (en) | Automatic variable-dividing encoder for depth layering | |
| CN111524207B (en) | Image generation method and device based on artificial intelligence and electronic equipment | |
| CN108961369A (en) | The method and apparatus for generating 3D animation | |
| US10515456B2 (en) | Synthesizing hair features in image content based on orientation data from user guidance | |
| CN111223057B (en) | Incremental focused image-to-image conversion method based on generation of countermeasure network | |
| US20220156987A1 (en) | Adaptive convolutions in neural networks | |
| CN106780363B (en) | Picture processing method and device and electronic equipment | |
| CN108549850A (en) | A kind of image-recognizing method and electronic equipment | |
| Chen et al. | A generative sketch model for human hair analysis and synthesis | |
| DE102022003003A1 (en) | Automatic photo editing using spoken instructions | |
| KR20220147545A (en) | Image editing model training method and image editing method | |
| CN111933253B (en) | Neural network-based marking point marking method and device for bone structure image | |
| CN111080746A (en) | Image processing method, image processing device, electronic equipment and storage medium | |
| CN117557686A (en) | Picture processing method, device, equipment and storage medium | |
| CN116843901A (en) | Medical image segmentation model training method and medical image segmentation method | |
| US20090040218A1 (en) | Fitting curves from one model to another | |
| CN117218300B (en) | Three-dimensional model construction method, three-dimensional model construction training method and device | |
| CN115526223A (en) | Score-based generative modeling in a potential space | |
| CN112862672A (en) | Bang generation method and device, computer equipment and storage medium | |
| US10762717B2 (en) | Blend shape system with dynamic partitioning | |
| CN117372644A (en) | Three-dimensional content generation method based on period implicit representation | |
| Bandyopadhyay et al. | Sketchinr: A first look into sketches as implicit neural representations |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |