CN105992011B - The calculation method and device of loop filtering - Google Patents
The calculation method and device of loop filtering Download PDFInfo
- Publication number
- CN105992011B CN105992011B CN201510088119.3A CN201510088119A CN105992011B CN 105992011 B CN105992011 B CN 105992011B CN 201510088119 A CN201510088119 A CN 201510088119A CN 105992011 B CN105992011 B CN 105992011B
- Authority
- CN
- China
- Prior art keywords
- mentioned
- coded treatment
- filtering
- blocking filtering
- processing
- 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.)
- Active
Links
Landscapes
- Image Processing (AREA)
- Compression Or Coding Systems Of Tv Signals (AREA)
Abstract
The present invention provides the calculation method and device of a kind of loop filtering.De-blocking filtering processing first is executed to the vertical boundary of coded treatment block, obtains multiple specified pixel points whereby.Here, the value of specified pixel point will not be changed when the horizontal boundary to coded treatment block executes de-blocking filtering processing.Then, edge offset is executed to above-mentioned specified pixel point to compare, and obtain comparison result.Later, de-blocking filtering processing is executed to the horizontal boundary of coded treatment block, and obtains result after de-blocking filtering.Also, according to result after above-mentioned de-blocking filtering and above-mentioned comparison result, adaptive offset filtering processing is sampled to coded treatment onblock executing.
Description
Technical field
The invention relates to a kind of video coding mechanisms, and a kind of calculation method in particular to loop filtering and
Device.
Background technique
High efficiency Video coding (High Efficiency Video Coding, abbreviation HEVC) is a kind of video compress mark
It is quasi-.HEVC then further decomposes brightness and chromatic component first by coding tree unit (Coding Tree Unit, CTU)
For code tree block (Coding Tree Block, CTB).Code tree block is usually 64 × 64,32 × 32 or 16 × 16, is led to
Normal piecemeal is bigger, and code efficiency is higher.There are two loop filters by HEVC, that is, de-blocking filtering device (deblocking
Filter, DBF) and the adaptive offset filter (Sample adaptive offset (SAO) filter) of sampling.It is general and
Speech, SAO filter needs to refer to eight pixel values of surrounding, therefore brings very big difficulty in design.
Summary of the invention
The present invention provides the calculation method and device of a kind of loop filtering, reduces the difficulty in design, and control logic is opposite
It is clear simple.
The calculation method of loop filtering of the invention, comprising: from image reconstruction data, sequentially take out coded treatment area
Block；De-blocking filtering (deblocking filter, DBF) processing is executed to the vertical boundary of coded treatment block, is obtained whereby
Multiple specified pixel points, wherein the value of specified pixel point will not when the horizontal boundary to coded treatment block executes DBF processing
It is changed；Edge offset (edge offset) is executed to above-mentioned specified pixel point to compare, and obtains comparison result；At coding
The horizontal boundary for managing block executes DBF processing, and obtains result after de-blocking filtering；And according to being tied after above-mentioned de-blocking filtering
Fruit and above-mentioned comparison result, to coded treatment onblock executing one sample adaptive offset (Sample adaptive offset,
SAO it) is filtered.
In one embodiment of this invention, DBF processing is executed in the vertical boundary to coded treatment block, is referred to whereby
In the step of determining pixel, with the horizontal boundary with coded treatment block at a distance of the position of the pixel of preset quantity as finger
Determine pixel.
In one embodiment of this invention, above-mentioned specified pixel point is parallel adjacent rows.And to specified pixel point
It executes in the step of edge offset compares, side is executed to more 2 × 2 specified pixel points respectively according to directionality scanning pattern
Edge offset compares and obtains comparison result.Here, directionality scanning pattern include horizontal sweep, vertical scanning, 135 degree diagonally sweep
It retouches and one of 45 degree of diagonal scans.
In one embodiment of this invention, above-mentioned after obtaining comparison result, it can also come according further to comparison result
Select corresponding deviant.
In one embodiment of this invention, above-mentioned to compare to specified pixel point execution edge offset, and obtain and compare knot
After fruit, storage comparison result to buffer.
In one embodiment of this invention, the complete DBF processing of current coded treatment onblock executing and above-mentioned sampling from
After adapting to offset filtering processing, from above-mentioned image reconstruction data, another coded treatment block is taken out sequentially to execute at DBF
Reason and SAO filtering processing.
In one embodiment of this invention, it holds taking out another coded treatment block from above-mentioned image reconstruction data
When row DBF processing, then the buffer for storing a preceding comparison result can be covered by the comparison result of another coded treatment block
Lid.
Loop filtering computing device of the invention, comprising: buffer；De-blocking filtering device, is coupled to buffer, wherein going
Blocking filtering device sequentially takes out coded treatment block from image reconstruction data, to coded treatment onblock executing DBF processing；
SAO filter is coupled to buffer and de-blocking filtering device, and wherein SAO filter is to the coded treatment area handled via DBF
Block executes SAO filtering processing.De-blocking filtering device executes DBF processing to the vertical boundary of coded treatment block, obtains whereby more
A specified pixel point, and the value of specified pixel point will not be changed when the horizontal boundary to coded treatment block executes DBF processing
Become.Also, de-blocking filtering device compares to specified pixel point execution edge offset, and after obtaining comparison result, it will compare
As a result it is stored in buffer, and DBF processing is executed to the horizontal boundary of coded treatment block, and is tied after obtaining de-blocking filtering
Fruit.According to result and comparison result after de-blocking filtering, SAO filter executes the coded treatment block handled via DBF
SAO filtering processing.
Based on above-mentioned, in the DBF processing for having executed vertical boundary, the comparison result of a part is calculated, first just for rear
It is used when continuous SAO filtering processing, processing sequence is still according to general DBF processing sequence, and control logic is relatively clear simple.
To make the foregoing features and advantages of the present invention clearer and more comprehensible, special embodiment below, and cooperate institute's accompanying drawings
It is described in detail below.
Detailed description of the invention
Fig. 1 is the block diagram according to the loop filtering computing device of one embodiment of the invention.
Fig. 2 is the block diagram according to the coded system of one embodiment of the invention.
Fig. 3 A~Fig. 3 C is the schematic diagram according to the DBF processing sequence of one embodiment of the invention.
Fig. 4 is that the directionality that the edge offset being filtered according to the SAO of one embodiment of the invention compares scans pattern
Schematic diagram.
Fig. 5 is the flow chart according to the calculation method of the loop filtering of one embodiment of the invention.
Fig. 6 is the schematic diagram handled according to one embodiment of the invention coded treatment onblock executing DBF.
Fig. 7 is the schematic diagram according to the execution loop filtering process of one embodiment of the invention.
Fig. 8 is the schematic diagram being filtered according to the SAO of one embodiment of the invention.
Description of symbols
100: loop filtering computing device
110: de-blocking filtering device
120:SAO filter
130: buffer
200: coded system
210: between picture/intra-frame prediction device
220: anti-discrete cosine converter
230: adder
240: image reconstruction processor
250: frame buffer
0~95: boundary
410~440: directionality scans pattern
6a, 6b, 6c, G, T1~T5: pixel
6d, C1, C2: comparison result
610,610 ': dotted line frame
A, B, D1~D4, S1~S2: block
P1, P2: comparator
VL: vertical boundary
S505~S525: each step of the calculation method of loop filtering
Specific embodiment
Fig. 1 is the block diagram according to the loop filtering computing device of one embodiment of the invention.Please refer to Fig. 1, loop filtering
Computing device 100 includes de-blocking filtering device (deblocking filter, DBF) 110, the adaptive offset (Sample of sampling
Adaptive offset, SAO) filter 120 and buffer 130.Buffer 130 is respectively coupled to de-blocking filtering device 110
And SAO filter 120.And SAO filter 120 is coupled to de-blocking filtering device 110.In high efficiency Video coding (High
Efficiency Video Coding, HEVC) in standard, de-blocking filtering device 110 and SAO filter 120 are for improving figure
As quality.
Fig. 2 is the block diagram according to the coded system of one embodiment of the invention.The coded system 200 of the present embodiment is only ring
Road filters the one of which application aspect of computing device 100, is not intended to limit the internal component of coded system 200 herein.In this reality
Apply in example, coded system 200 include picture between/intra-frame prediction device 210, anti-discrete cosine converter 220, adder 230, figure
As reconstruction processor 240, loop filtering computing device 100 and frame buffer (frame buffer) 250.
Between picture/intra-frame prediction device 210 to input data execute both inter-picture prediction and intra-frame prediction extremely
Few one.Anti-discrete cosine converter (inverse discrete cosine transformer) 220 is to input data
Execute anti-discrete cosine conversion (Inverse Discrete Cosine Transform, IDCT).Adder 230 is coupled to picture
Between face/intra-frame prediction device 210, anti-discrete cosine converter 220 and image reconstruction processor 240, between picture/
The output data of 220 the two of intra-frame prediction device 210 and anti-discrete cosine converter executes arithmetical operation.Image reconstruction processor
The output of 240 pairs of adders 230 executes reconstruction (reconstruct) movement.That is, input image via between picture/picture in it is pre-
Device 210 and anti-discrete cosine converter 220 are surveyed, then is sent to image reconstruction processor 240 via adder 230 to obtain
Image reconstruction (reconstructed) data.Frame buffer 250 is a video output device, will be calculated via loop filtering
The data that device 100 exports are exported to display units such as screens.
De-blocking filtering device 110 is that a kind of reduce generates the filter of visually flaw in block border.Here, removing block
Filter 110 is to take out a coded treatment block, sequentially to go to coded treatment onblock executing from image reconstruction data
Blocking filtering handles (abbreviation DBF processing).Here, coded treatment block be, for example, coding tree unit (Coding Tree Unit,
CTU).Coding tree unit is the basic coding unit of HEVC.HEVC supports the CTU size of 16 × 16 to 64 × 64 pixels.Input
Image can be first cut into the coding tree units of equal sizes before being encoded input coding device is encoded again.Assuming that input image
Using the format of YUV, then a coding tree unit includes code tree block (the Coding Tree of a brightness (Luma)
Block, CTB), the code tree block and syntactic element (Syntax Element) of two colorations (Chroma).
For example, Fig. 3 A~Fig. 3 C is the schematic diagram according to the DBF processing sequence of one embodiment of the invention.Here, figure
3A show the code tree block of 32 × 32 brightness, is respectively 16 × 16 chroma blue and red color shown in Fig. 3 B and Fig. 3 C
The code tree block of degree.Here, each 8 × 8 code tree block can be to its left boundary (belonging to vertical boundary) and side above
Boundary's (belonging to horizontal boundary) executes filtering processing.In DBF processing, DBF processing first is executed to vertical boundary and then to level
Boundary executes DBF processing.The number of Fig. 3 A~Fig. 3 C is to execute the sequence of filtering.In addition, if the boundary of 8 × 8 blocks is shadow
The boundary of picture does not execute DBF processing then.
SAO filter 120 executes SAO filtering processing to via DBF treated coded treatment block.SAO filtering processing
Different modes of deflection can be used according to demand.In general, SAO filtering processing will use edge offset (edge
Offset) mode or with offset (band offset) mode.Edge offset mode is based on more current sample and neighbouring sample
This, and be based on sample value with modes of deflection.
In the present embodiment, loop filtering computing device 100 further includes buffer 130.Buffer 130 is configured to temporarily store following
Comparison result, that is, after de-blocking filtering device 110 executes DBF processing to the vertical boundary of a coded treatment block, to specified picture
Vegetarian refreshments executes edge offset comparison result more obtained.And SAO filter 120 can according to result after de-blocking filtering and
The comparison result of buffer 130 executes SAO filtering processing to the coded treatment block handled via DBF.
The design of SAO filter 120 is and to reduce oscillation effect to improve image quality.Here, SAO is filtered
Use edge offset mode.That is, SAO filter 120 is by comparing the adjacent pixel value of target pixel value, to object pixel
Value is classified, and applies a deviant according to classification later.For example, based on two pictures adjacent with target pixel value are compared
Element value, classifies to target pixel value.In SAO filtering processing, the comparison result that edge offset compares includes five classification,
That is, minimum, two kinds of edges, maximum values, or both are not.
For example, Fig. 4 is that the directionality that the edge offset being filtered according to the SAO of one embodiment of the invention compares is swept
Retouch the schematic diagram of pattern.Here, SAO filtering processing provides 4 kinds of directionality scanning patterns 410~440, respectively level is swept
It retouches, vertical scanning, 135 degree of diagonal scans and 45 degree of diagonal scans.Each coded treatment block will use identical one kind
Directionality scans pattern.For example, it is assumed that having selected directionality to scan pattern 410, then at each coding of image reconstruction data
Reason block all will use directionality scanning pattern 410 and compare to execute edge offset.Wherein, " c " represents current sample, " a "
" b " then represents neighbouring sample.The rule that edge offset compares is as shown in table 1.SAO filter 120 is according to comparison result
(that is, classification 0~4) gives a corresponding deviant.
Table 1
| Classification | Edge offset compares |
| 1 | c<a&&c<b |
| 2 | (c < a&&c==b) | | (c==a&&c <b) |
| 3 | (c > a&&c==b) | | (c==a&&c > b) |
| 4 | c>a&&c>b |
| 0 | It is all no above |
Above-mentioned each component can be the hardware or chip being made of multiple digital logic gates, can also be by programming language
The software write and driven by the arithmetic elements such as processor or processing chip.
It is beneath to arrange in pairs or groups above-mentioned component to illustrate the calculation method of loop filtering.Fig. 5 is the ring according to one embodiment of the invention
The flow chart of the calculation method of road filtering.Firstly, in step S505, through de-blocking filtering device 110 from image reconstruction data
In, sequentially take out a coded treatment block.
Then, in step S510, DBF is executed through vertical boundary of the de-blocking filtering device 110 to coded treatment block
Processing, obtains multiple specified pixel points whereby.Here, the value of specified pixel point is executed in the horizontal boundary to coded treatment block
It will not be changed when DBF processing.Due at most can only change with boundary both sides in preset quantity when executing DBF processing
Pixel value.Therefore after the DBF processing for having executed vertical boundary, can obtain will not be by when horizontal boundary executes DBF processing
The specified pixel point of change.That is, being made with the horizontal boundary with coded treatment block at a distance of the position of the pixel of preset quantity
For specified pixel point.
Fig. 6 is the schematic diagram handled according to one embodiment of the invention coded treatment onblock executing DBF.Please refer to Fig. 6, area
Block A indicates the block that DBF processing is executed to vertical boundary VL.In this, it is assumed that at most can only change boundary two executing DBF processing
The value of 3 pixels on side, therefore after executing DBF processing to vertical boundary VL, it can be learnt that the 4th pixel value of horizontal boundary
For specified pixel point.It is represented with the hollow pixel 6a indicated when executing DBF processing to horizontal boundary, which can be changed.
It is represented with the pixel 6b that solid black indicates when executing DBF processing to horizontal boundary, which will not be changed.With horizontal line
The pixel 6c of expression represents the point and will not be changed being filtered to vertical boundary or horizontal boundary.
Then, in step S515, de-blocking filtering device 110 executes edge offset to specified pixel point and compares, and then obtains
Obtain comparison result.Later, in step S520, de-blocking filtering device 110 executes at DBF the horizontal boundary of coded treatment block
Reason, and obtain result (that is, the pixel value handled by the DBF of vertical boundary and horizontal boundary) after de-blocking filtering.
Here, specified pixel point is parallel adjacent rows.As shown in fig. 6, parallel adjacent rows pixel includes more
A pixel 6b, 6c.From the point of view of dotted line frame 610 is enlarged into dotted line frame 610 ', de-blocking filtering device 110 can be respectively to more 2 × 2
Specified pixel point compare to execute edge offset, select deviant according to comparison result later.Here, to 2 × 2 it is specified
Pixel executes edge offset comparison result more obtained there are three types of situation, i.e., less than, greater than or equal to.
Compare here, providing 4 kinds of directionality scanning patterns and executing edge offset to specified pixel point, above-mentioned directionality is swept
Retouching pattern includes vertical scanning (two pixels are compared above and below selecting), horizontal sweep (the pixels click-through of selection left and right two
Row compares), 135 degree diagonal scans (two pixels of 135 degree of selection diagonally are compared) and 45 degree of diagonal scans (select
45 degree of two diagonal pixels are compared).Each group 2 × 2 of specified pixel point will use a kind of identical directionality and sweep
Retouch pattern.Comparing the relationship between two pixels is to be less than, be greater than or equal to.Later, according to comparison result (that is, being less than, greatly
In or be equal to) give a corresponding deviant.Referring to the dotted line frame 610 ' of Fig. 6,2 × 2 finger is indicated with open squares
Determine the comparison result 6d of pixel.And these comparison results 6d can be stored to buffer 130.
Fig. 7 is the schematic diagram according to the execution loop filtering process of one embodiment of the invention.In the present embodiment, with 16 ×
16 coded treatment block CTB is illustrated.In the present embodiment, facilitate explanation to ask, only show and execute DBF processing
Block D1~D4 and block S1~the S2 for executing SAO filtering processing.
De-blocking filtering device 110 has executed vertical boundary (vertical boundary 0~11) to current coded treatment block CTB
After DBF processing, specified pixel point (including the pixel 6b indicated with solid black and the pixel 6c indicated with horizontal line)
Value would not be changed.In detail, de-blocking filtering device 110 executes DBF processing to vertical boundary 0 in block D1, in area
DBF processing is executed to vertical boundary 1 in block D2, DBF processing is executed to vertical boundary 2 in block D3, to vertical in block D4
Straight boundary 3 executes DBF processing etc., and so on.To vertical boundary 0~11 execute DBF processing after, with horizontal boundary apart
The position of 3 pixels is as specified pixel point (that is, the pixel 6b indicated with solid black and the picture indicated with horizontal line
Vegetarian refreshments 6c).Whereby, de-blocking filtering device 110 obtains the ratio of 2 × 2 specified pixel point according to the filter result of vertical boundary
Compared with result 6d, indicated in Fig. 7 with open squares to indicate comparison result 6d (non-pixel).
In Fig. 7, when handling the coded treatment block CTB DBF for executing vertical boundary, due to coded treatment block CTB
The vertical boundary 16~19 of rightmost needs to use the value of next coded treatment block CTB, therefore at current coding
DBF processing will not be executed to vertical boundary 16~19 by managing in block CTB.
In addition, when handling the coded treatment block CTB DBF for executing horizontal boundary, although horizontal boundary 32,40 is not in mesh
Within the scope of preceding coded treatment block CTB, but DBF processing can be executed to horizontal boundary 32,40.And at current coding
The horizontal boundary 36,44 in block CTB is managed, since it is desired that using the value of next coded treatment block CTB, therefore current
Coded treatment block CTB in will not to horizontal boundary 36,44 execute DBF processing.
Fig. 5 is returned to, in step S525, SAO filter 120 is according to result after de-blocking filtering and comparison result (storage
It is stored in buffer 130), coded treatment onblock executing SAO is filtered.In detail, it after executing the step S520, goes
Blocking filtering device 110, which is understood, be sent to SAO filter 120 for result after de-blocking filtering, meanwhile, SAO filter 120 is understood from temporary
The comparison result previously stored is read out into use in device 130.
In Fig. 7, after whole vertical boundaries to coded treatment block CTB have executed DBF processing, de-blocking filtering device
110 can execute one by one DBF processing to horizontal boundary.And after the DBF processing for having executed a horizontal boundary, de-blocking filtering
Result after the de-blocking filtering of this horizontal boundary can be sent to SAO filter 120 by device 110.And SAO filter 120
According to the comparison result in result and buffer 130 after the de-blocking filtering received, Lai Zhihang SAO filtering processing.For example, going
Blocking filtering device 110, will be by the de-blocking filtering obtained of horizontal boundary 32 after the DBF processing for having executed horizontal boundary 32
Result is sent to SAO filter 120 afterwards, and does not have to just pass until whole horizontal boundaries (32~35) has all executed DBF processing
Result is sent to after blocking filtering to SAO filter 120.Similarly, DBF processing and SAO are executed to block S2 (and other blocks)
Filtering processing, and so on.
Fig. 8 is the schematic diagram being filtered according to the SAO of one embodiment of the invention.Fig. 8 is shown with the horizontal sides in Fig. 7
Partial schematic diagram at boundary 40.In fig. 8, with the comparator of double circle (bullseye) presentation-entity with one heart, these comparators are used
To compare via DBF neighbouring four pixels around treated, and then it can get a comparison result.Also, with open squares
Indicate the comparison result being stored in buffer 130, that is, the comparison result obtained in DBF processing.In addition, being come with circle
Indicate each pixel.
It is beneath to be illustrated for executing SAO filtering processing to pixel G.Before executing SAO filtering processing, own
Pixel has all executed DBF processing.That is, the value of pixel G and pixel T1~T5 are result after de-blocking filtering.Comparator P1
Comparison result P1 ' is obtained according to pixel G and pixel T1~T3, comparator P2 is obtained according to pixel G and pixel T3~T5
Obtain comparison result P2 '.And SAO filter 120 can be according to comparison result C1, the C2 and comparator stored in buffer 130
P1, P2 comparison result P1 ' obtained, P2 ' Lai Zhihang SAO filtering processing.And so on, it can get block B all pixels point
Output after being filtered via SAO.
So, it above are only a kind of selection of SAO filtering processing.SAO filter 120 can also be stored up according in buffer 130
Comparison result C1, C2 for depositing and result executes at SAO filtering after de-blocking filtering of other pixels T1~T5 around it
Reason.
And after current coded treatment block CTB has executed DBF processing and SAO filtering processing, de-blocking filtering
Device 110 can sequentially take out next coded treatment block from image reconstruction data to execute at DBF processing and SAO filtering
Reason, until whole image, which rebuilds data, all completes the calculating of loop filtering.
In addition, when taking out next coded treatment block from image reconstruction data to execute DBF processing, then it is preceding primary
The comparison result for being stored in buffer 130 can be covered by the comparison result of this coded treatment block, temporary to save whereby
The use space of device 130.
In conclusion calculating the comparison result of a part executing DBF processing Shi Huixian to vertical boundary, and will compare
As a result it stores into buffer, when the DBF processing for having executed horizontal boundary enters SAO filtering processing, is read from buffer
Comparison result is taken to execute SAO filtering processing.Accordingly, it is handled by the DBF for having executed vertical boundary, just first calculates and compare
As a result, for 32 × 32 coded treatment block, it is only necessary to which the buffer space of 4 × 32 × 2=256 bit (bit) is stored up
Comparison result is deposited, the expense of hardware can be saved.In addition, in above-mentioned loop filtering calculation method, processing sequence be still according to
DBF processing sequence, therefore when to execute the sharing of multi-format (multi format) in the future, hardware resource easy to accomplish melts
It closes.
Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention., rather than its limitations；To the greatest extent
Pipe present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that: its according to
So be possible to modify the technical solutions described in the foregoing embodiments, or to some or all of the technical features into
Row equivalent replacement；And these are modified or replaceed, various embodiments of the present invention technology that it does not separate the essence of the corresponding technical solution
The range of scheme.
Claims (10)
1. a kind of calculation method of loop filtering characterized by comprising
From in an image reconstruction data, a coded treatment block is sequentially taken out；
The processing of one de-blocking filtering is executed to the vertical boundary of above-mentioned coded treatment block, obtains multiple specified pixel points whereby,
Wherein the value of above-mentioned specified pixel point is when the horizontal boundary to above-mentioned coded treatment block executes the processing of above-mentioned de-blocking filtering
It will not be changed by filtering；
To above-mentioned specified pixel point execute an edge deviation ratio compared with, and obtain a comparison result；
Above-mentioned de-blocking filtering processing is executed to the horizontal boundary of above-mentioned coded treatment block, and is tied after obtaining a de-blocking filtering
Fruit；And
According to result after above-mentioned de-blocking filtering and above-mentioned comparison result, above-mentioned coded treatment onblock executing one is sampled adaptive
Filtering processing should be deviated.
2. the calculation method of loop filtering as described in claim 1, which is characterized in that hang down to above-mentioned coded treatment block
The step of straight boundary executes the processing of above-mentioned de-blocking filtering, obtains above-mentioned specified pixel point whereby include:
With the above-mentioned horizontal boundary with above-mentioned coded treatment block at a distance of the position of the pixel of a preset quantity as above-mentioned
Specified pixel point.
3. the calculation method of loop filtering as described in claim 1, which is characterized in that above-mentioned specified pixel point is parallel phase
Adjacent two rows, and the step of comparing the above-mentioned edge offset of above-mentioned specified pixel point execution includes:
According to directionality scanning pattern respectively to more 2 × 2 above-mentioned specified pixel points execute above-mentioned edge offset compare and
Obtain above-mentioned comparison result, wherein above-mentioned directionality scanning pattern include horizontal sweep, vertical scanning, 135 degree of diagonal scans with
And one of 45 degree of diagonal scans.
4. the calculation method of loop filtering as claimed in claim 3, which is characterized in that after obtaining above-mentioned comparison result,
It further includes:
Corresponding deviant is selected according to above-mentioned comparison result.
5. the calculation method of loop filtering as described in claim 1, which is characterized in that executed to above-mentioned specified pixel point
Edge offset is stated to compare, and after the step of obtaining above-mentioned comparison result, it further includes:
Above-mentioned comparison result is stored to a buffer.
6. the calculation method of loop filtering as described in claim 1, which is characterized in that in current above-mentioned coded treatment block
It has executed above-mentioned de-blocking filtering processing and above-mentioned sampling adaptively deviates after being filtered, from above-mentioned image reconstruction data
In, take out another coded treatment block sequentially to execute above-mentioned de-blocking filtering processing and above-mentioned sampling and adaptively deviate filtering
Processing.
7. the calculation method of loop filtering as claimed in claim 6, which is characterized in that taken from above-mentioned image reconstruction data
When another above-mentioned coded treatment block is handled out to execute above-mentioned de-blocking filtering, then primary above-mentioned comparison result before storing
Buffer can be covered by the comparison result of another above-mentioned coded treatment block.
8. a kind of loop filtering computing device characterized by comprising
One buffer；
One de-blocking filtering device, is coupled to the buffer, wherein above-mentioned de-blocking filtering device is from an image reconstruction data, sequentially
A coded treatment block is taken out, to the processing of above-mentioned one de-blocking filtering of coded treatment onblock executing；
The one adaptive offset filter of sampling, is coupled to the buffer and above-mentioned de-blocking filtering device, wherein above-mentioned sampling is adaptive
It answers offset filter to the above-mentioned coded treatment block handled via above-mentioned de-blocking filtering, executes the adaptive offset filter of a sampling
Wave processing；
Wherein, above-mentioned de-blocking filtering device executes above-mentioned de-blocking filtering to the vertical boundary of above-mentioned coded treatment block and handles,
Multiple specified pixel points are obtained whereby, wherein the value of above-mentioned specified pixel point is held in the horizontal boundary to above-mentioned coded treatment block
It will not be changed by filtering when row above-mentioned de-blocking filtering processing, and
Above-mentioned de-blocking filtering device to above-mentioned specified pixel point execute an edge deviation ratio compared with, and obtain a comparison result it
Afterwards, which is stored in the buffer, and removes block to the execution of the horizontal boundary of above-mentioned coded treatment block is above-mentioned
Filtering processing, and obtain a de-blocking filtering after as a result,
According to result after above-mentioned de-blocking filtering and above-mentioned comparison result, the above-mentioned adaptive offset filter of sampling is to via upper
The above-mentioned coded treatment block for stating de-blocking filtering processing, executes above-mentioned sampling and adaptively deviates filtering processing.
9. loop filtering computing device as claimed in claim 8, which is characterized in that above-mentioned de-blocking filtering device is to above-mentioned volume
After the vertical boundary of code processing block executes above-mentioned de-blocking filtering processing, with the above-mentioned horizontal sides with above-mentioned coded treatment block
Boundary is at a distance of the position of the pixel of a preset quantity as above-mentioned specified pixel point.
10. loop filtering computing device as claimed in claim 8, which is characterized in that above-mentioned specified pixel point is parallel phase
Adjacent two rows, and above-mentioned de-blocking filtering device respectively holds more 2 × 2 above-mentioned specified pixel points according to directionality scanning pattern
The above-mentioned edge offset of row compares and obtains above-mentioned comparison result, and selectes corresponding deviant according to above-mentioned comparison result,
In above-mentioned directionality scanning pattern include horizontal sweep, vertical scanning, 135 degree of diagonal scans and 45 degree of diagonal scans wherein it
One.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510088119.3A CN105992011B (en) | 2015-02-26 | 2015-02-26 | The calculation method and device of loop filtering |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510088119.3A CN105992011B (en) | 2015-02-26 | 2015-02-26 | The calculation method and device of loop filtering |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105992011A CN105992011A (en) | 2016-10-05 |
| CN105992011B true CN105992011B (en) | 2019-04-09 |
Family
ID=57039082
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510088119.3A Active CN105992011B (en) | 2015-02-26 | 2015-02-26 | The calculation method and device of loop filtering |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105992011B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113891075B (en) * | 2020-07-03 | 2023-02-28 | 杭州海康威视数字技术股份有限公司 | Filtering processing method and device |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102860005A (en) * | 2010-02-05 | 2013-01-02 | 瑞典爱立信有限公司 | De-blocking filtering control |
| CN103051892A (en) * | 2011-10-14 | 2013-04-17 | 联发科技股份有限公司 | Method and apparatus for in-loop filtering |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| PT3001682T (en) * | 2011-03-01 | 2017-05-18 | ERICSSON TELEFON AB L M (publ) | Deblocking filtering control |
| WO2013053324A1 (en) * | 2011-10-14 | 2013-04-18 | Mediatek Inc. | Method and apparatus for loop filtering |
-
2015
- 2015-02-26 CN CN201510088119.3A patent/CN105992011B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102860005A (en) * | 2010-02-05 | 2013-01-02 | 瑞典爱立信有限公司 | De-blocking filtering control |
| CN103051892A (en) * | 2011-10-14 | 2013-04-17 | 联发科技股份有限公司 | Method and apparatus for in-loop filtering |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105992011A (en) | 2016-10-05 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108184129B (en) | Video coding and decoding method and device and neural network for image filtering | |
| TWI545526B (en) | Decoder, decoding method, and related encoder, method for encoding, and digital storage medium | |
| CN103155557B (en) | Adaptive loop filter method and apparatus based on subregion basis | |
| CN102708542B (en) | Generating edge masks for a deblocking filter | |
| CA3105023A1 (en) | Intra-frame prediction method and device | |
| CN105491376B (en) | Image coding and decoding method and device | |
| CN109842799B (en) | Intra-frame prediction method and device of color components and computer equipment | |
| CN110892719A (en) | Image encoding/decoding method and apparatus | |
| CN107197258A (en) | Loop circuit filtering method and its equipment | |
| JP2017508415A (en) | Image encoding / decoding method and apparatus | |
| CN103947208A (en) | Method and apparatus for reduction of deblocking filter | |
| CN104754362B (en) | Image compression method using fine-divided block matching | |
| CN103051892B (en) | Embedded loop filter method and embedded loop filter | |
| TWI833073B (en) | Coding using intra-prediction | |
| CN104521234B (en) | Merge the method for processing video frequency and device for going block processes and sampling adaptive migration processing | |
| CN110971897B (en) | Method, apparatus and system for encoding and decoding intra prediction mode of chrominance component | |
| JP6794364B2 (en) | Image coding method and device and image decoding method and device | |
| Wang et al. | Parallel in-loop filtering in HEVC encoder on GPU | |
| CN107040778A (en) | Loop circuit filtering method and loop filter | |
| US11991399B2 (en) | Apparatus and method for de-blocking filtering | |
| CN105992011B (en) | The calculation method and device of loop filtering | |
| CN106303535B (en) | Image compression method and device with reference pixels taken from different-degree reconstruction pixels | |
| JP6768756B2 (en) | Video processing equipment and video processing method | |
| CA3104069A1 (en) | Video coding comprising uniform tile split with remainder | |
| WO2016124154A1 (en) | Image coding method and apparatus, and image decoding method and apparatus |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |