CN106375762A - Reference frame data compression method and apparatus - Google Patents
Reference frame data compression method and apparatus Download PDFInfo
- Publication number
- CN106375762A CN106375762A CN201510435316.8A CN201510435316A CN106375762A CN 106375762 A CN106375762 A CN 106375762A CN 201510435316 A CN201510435316 A CN 201510435316A CN 106375762 A CN106375762 A CN 106375762A
- Authority
- CN
- China
- Prior art keywords
- group
- data block
- pixel
- predictive mode
- equations
- 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
Abstract
The invention relates to the field of image processing and discloses a reference frame data compression method and apparatus. The compression method comprises the following steps: segmenting a reference frame into multiple data blocks with predetermined sizes; for each data block, calculating compression rates when various candidate prediction modes are applied, and selecting a prediction mode with a highest compression rate for compressing the data blocks, wherein each candidate prediction mode only employs each pixel point in the corresponding data block for in-block prediction, each candidate prediction mode segments the data blocks into multiple groups, each group comprises multiple pixel points, and each candidate prediction mode comprises a prediction mode in which the pixel points in the groups are distributed according to a 45-degree or 135-degree direction. According to the invention, the prediction mode corresponding to the highest compression rate can be selected for individually compressing the data blocks, such that the compression rate and the compression success rate of the reference frame are effectively improved.
Description
Technical field
The present invention relates to image processing field, particularly to reference frame data compress technique.
Background technology
With the continuous development of video encoding standard, all kinds of coding standards, such as a kind of h.264 (video
Codec standard), h.265 (a kind of video encoding standard), vp9 (a kind of video compression standard)
Deng, the encoded picture size supported is increasing, high definition coding, 4k coding become homely food.
The change of dimension of picture directly results in greatly increasing considerably of reference frame data, the readwrite bandwidth of encoder chip
Also followed by increase, in the nervous encoder system of some bandwidth, the performance to encoder for the readwrite bandwidth
To have a direct impact.
At present, reference frame compression method can be simply divided into two classes: a class is lossy compression method, and a class is
Lossless compress.With respect to lossless compress, lossy compression method has more preferable compression ratio, but this can reduce coding
The quality of image, uses lossless compress or damage for this most of encoder and mutually ties with lossless compress
The method closed.But, current compression method still has that compression ratio is low, coding is dumb, transport is multiple
Miscellaneous degree is high, the problems such as coding is complicated, develops a kind of high reference frame compression method of compression ratio for this to coding
Most important for device.
Content of the invention
It is an object of the invention to provide a kind of reference frame data compression method and its device, it is right to can select
The predictive mode answering highest compression ratio is individually compressed to data block, thus effectively improving the pressure of reference frame
Shrinkage and compression success rate.
For solving above-mentioned technical problem, embodiments of the present invention disclose a kind of reference frame data compression side
Method, comprises the following steps:
Reference frame is divided into the data block of multiple predefined sizes;
To each data block, calculate the compression ratio during predictive mode applying various candidates respectively, and select
Select compression ratio highest predictive mode this data block is compressed;Wherein, the prediction mould of each candidate
Formula only carries out block interior prediction using each pixel in notebook data block, and the predictive mode of each candidate is respectively by number
It is divided into multiple groups according to block, each group includes multiple pixels, and the predictive mode of each candidate is included in group
Pixel presses the predictive mode of 45 degree or 135 degree directional spreding.
Embodiments of the present invention also disclose a kind of reference frame data compressor, comprising:
Cutting unit, for being divided into the data block of multiple predefined sizes by reference frame;
Compression unit, for each data block, when calculating the predictive mode applying various candidates respectively
Compression ratio, and select compression ratio highest predictive mode that this data block is compressed;
Wherein, the predictive mode of each candidate only carries out block interior prediction using each pixel in notebook data block,
Data block is divided into multiple groups by the predictive mode of each candidate respectively, and each group includes multiple pixels, respectively
The predictive mode of candidate includes in group pixel by the predictive mode of 45 degree or 135 degree directional spreding.
Compared with prior art, the main distinction and its effect are embodiment of the present invention:
By compression ratio under different predictive modes for each data block of independent calculating reference frame, can select
The predictive mode of corresponding highest compression ratio is individually compressed to data block, thus effectively improving reference frame
Compression ratio and compression success rate.
Further, the arithmetic path of decompression can be reduced, thus reducing the pressure of decompression engine, with
As a example the data block of 8x8, initial predicted pixel to the coordinate being (0,0) from coordinate is (7,7)
Pixel needs the adder through 14 grades, and the requirement to compression engine is higher, and works as initial prediction pixel
Coordinate when being changed to (3,3)/(3,4)/(4,3)/(4,4), longest path is from (3,3)
Only need to be through 8 grades of adders to (7,7), this all has side to optimizing circuit sequence with arrangement flowing water very much
Help.
Brief description
Fig. 1 is a kind of schematic flow sheet of reference frame data compression method in first embodiment of the invention;
Fig. 2 a and 2b is that the packet of data block under a kind of predictive mode in first embodiment of the invention is illustrated
Figure;
Fig. 3 a and 3b is that the packet of data block under a kind of predictive mode in first embodiment of the invention is illustrated
Figure;
Fig. 4 a and 4b is that the packet of data block under a kind of predictive mode in first embodiment of the invention is illustrated
Figure;
Fig. 5 a and 5b is that the packet of data block under a kind of predictive mode in first embodiment of the invention is illustrated
Figure;
Fig. 6 is a kind of structural representation of reference frame data compressor in third embodiment of the invention.
Specific embodiment
In the following description, in order that reader more fully understand the application and propose many technology thin
Section.But, even if it will be understood by those skilled in the art that there is no these ins and outs and be based on
The many variations of following embodiment and modification are it is also possible to realize the required guarantor of each claim of the application
The technical scheme of shield.
For making the object, technical solutions and advantages of the present invention clearer, below in conjunction with accompanying drawing to this
Bright embodiment is described in further detail.
First embodiment of the invention is related to a kind of reference frame data compression method.Fig. 1 is this reference frame number
Schematic flow sheet according to compression method.
Specifically, as shown in figure 1, this reference frame data compression method comprises the following steps:
In a step 101, reference frame is divided into the data block of multiple predefined sizes.
Then into step 102, to each data block, calculate the prediction mould applying various candidates respectively
Compression ratio during formula, and select compression ratio highest predictive mode that this data block is compressed.Its
In, the predictive mode of each candidate only carries out block interior prediction, each candidate using each pixel in notebook data block
Predictive mode respectively data block is divided into multiple groups, each group includes multiple pixels, each candidate's
Predictive mode includes in group pixel by the predictive mode of 45 degree or 135 degree directional spreding.
Hereafter, process ends.
Additionally, in a preference, in a step 102, multiple groups of the segmentation of data block predicted pattern
Including:
Pixel edge in one group of first kind group being made up of the pixel on 135 degree of diagonal and at least two groups groups
45 degree of directional spreding are in the Equations of The Second Kind group of 135 degree of diagonal both sides, and the number of pixels in first kind group is
/ 2nd of number of pixels in Equations of The Second Kind group.Or one group is made up of the pixel on 45 degree of diagonal
Pixel in first kind group and at least two groups groups is along 135 degree of directional spreding in the of 45 degree of diagonal both sides
Two class groups, and the number of pixels in first kind group is 1/2nd of number of pixels in Equations of The Second Kind group.
Specifically, as shown in figures 2 a and 2b, taking 8 × 8 data block as a example, the data shown in Fig. 2 a
The first kind group of block is made up of the pixel on 135 degree of diagonal, and wherein initial predicted pixel (black) is
4th pixel on diagonal, first kind group is divided into two first kind subgroups, four groups of Equations of The Second Kind group edges by it
45 degree of directional spreding, being divided into totally 8 Equations of The Second Kind subgroups by the pixel on 135 degree of diagonal respectively (needs
It is noted that in accompanying drawing of the present invention, color identical pixel forms one group).Shown in Fig. 2 b
The first kind group of data block is made up of the pixel on 45 degree of diagonal, and wherein, initial predicted pixel is right
4th pixel on linea angulata, first kind group is divided into two first kind subgroups by it, and four groups of Equations of The Second Kind groups are along 135
Degree directional spreding, is divided into totally 8 Equations of The Second Kind subgroups by the pixel on 45 degree of diagonal respectively.
Additionally, in another preference, also including the first rotation predictive mode in the predictive mode of each candidate,
Data block is included by multiple groups of the first rotation predictive mode segmentation:
One group of first kind group and at least two being made up of the pixel in a line in this data block center two row
The Equations of The Second Kind group that group is made up of the adjacent two row pixels of this data block.Wherein, this predictive mode is initially pre-
Survey pixel and be located at the row that first kind group is located, first kind group is divided into two first kind by initial predicted pixel
Group, the prediction direction of two first kind subgroups is contrary, and the row that each Equations of The Second Kind group is located by first kind group divides
Become two Equations of The Second Kind subgroups, the prediction direction belonging to two Equations of The Second Kind subgroups of an Equations of The Second Kind group together is mutually vertical
Directly, the prediction direction and between subgroup is along changing clockwise or counterclockwise.Specifically, as Fig. 3 a institute
Show, taking 8 × 8 data block as a example, the first kind group of the data block shown in Fig. 3 a is by 7 in fourth line
Individual pixel composition, initial predicted pixel coordinate is (4,4), i.e. fourth line the 4th row, by first kind component
Become two groups of first kind subgroups, wherein organize a1Prediction direction to the left, a2Prediction direction to the right.Data block
Adjacent column forms Equations of The Second Kind group, and the Equations of The Second Kind group of the wherein the 1st row and the 2nd row composition is divided into b1And b2
Two Equations of The Second Kind subgroups, prediction direction respectively upwards and to the left, the like, prediction direction is with a1→
b1+c1→d1+e1→a2→d2+e2→c2+b2Order rotate clockwise change.
Furthermore, it is to be understood that in the other embodiment of the present invention, being divided by the first rotation predictive mode
In data block after group the prediction direction of each group can also from different in Fig. 3 a, for example, prediction direction
With a1→b1+c1→d1→e1→a2→d2→e2→c2→b2Order along becoming clockwise or counterclockwise
Change, etc..
Additionally, in another preference, the predictive mode of each candidate includes the second rotation predictive mode,
Data block is included by multiple groups of the second rotation predictive mode segmentation:
One group arranged by this data block center two in string on the first kind group and at least two that forms of pixel
The Equations of The Second Kind group that group is made up of the adjacent rows pixel of this data block.Wherein, this predictive mode is initially pre-
Survey pixel and be located at the row that first kind group is located, first kind group is divided into two first kind by initial predicted pixel
Group, the prediction direction of two first kind subgroups is contrary, and the row that each Equations of The Second Kind group is located by first kind group divide
Become two Equations of The Second Kind subgroups, the prediction direction belonging to two Equations of The Second Kind subgroups of an Equations of The Second Kind group together is mutually vertical
Directly, the prediction direction and between subgroup is along changing clockwise or counterclockwise.Specifically, as Fig. 3 b institute
Show, taking 8 × 8 data block as a example, similar with Fig. 3 a, difference is the picture by the 4th row for the first kind group
Element composition, Equations of The Second Kind group is made up of adjacent lines.Wherein, prediction direction is with a3→b3+c3→d3→e3→a4
→d4+e4→c4+b4Order along rotate counterclockwise change.In the same manner, it is grouped by the first rotation predictive mode
In data block afterwards the prediction direction of each group can also from different in Fig. 3 b.
Furthermore, it is to be understood that the multiformity based on image texture change, in other embodiment party of the present invention
In formula, candidate modes can also include other predictive modes, and such as horizontal prediction mode is (as Fig. 4 a
With shown in Fig. 4 b), vertical predictive mode (as shown in figure 5 a and 5b) etc..Wherein, predict
Direction as shown by arrows, represents that the horizontal line in the grid of pixel represents the meansigma methodss taking two neighboring pixel,
Take the meansigma methodss of the two neighboring pixel of same row in fig. 4b, take the two neighboring picture of same a line in figure 5b
The meansigma methodss of element.
Preferably, the data block in the application includes nxn pixel, and wherein n is just whole more than 4
Number.The initial predicted pixel of the predictive mode of each candidate is one of pixel positioned at data block center.
Using the pixel at center as initial predicted pixel, the arithmetic path of decompression can be reduced, thus reducing
The pressure of decompression engine, as shown in Figs. 4a and 4b, taking the data block of 8x8 as a example, from coordinate be (0,
0) pixel that initial predicted pixel is (7,7) to coordinate needs the adder through 14 grades, to pressure
The requirement of contracting engine is higher, and the coordinate working as initial prediction pixel be changed to (3,3)/(3,4)/(4,
3), when/(4,4), longest path arrives (7,7) from (3,3) only need to be through 8 grades of adders, this
All very helpful with arrangement flowing water to optimizing circuit sequence.
Furthermore, it is to be understood that it is also possible to not adopt center pixel in the other embodiment of the application
As initial predicted pixel, here is not limited.
In another preference, above-mentioned steps 102 include following sub-step:
Difference between the predictive value of each pixel and actual value in data block is calculated according to current prediction mode
Value.Determine the volume of respective sets pixel according to the difference of maximum absolute value in each group that data block is divided into
Pattern.Calculate the bit number needed for each group being encoded using the corresponding encoded pattern that each group determines,
And compression ratio under current prediction mode for the data block is determined based on the bit number calculating gained.
Additionally, the difference of maximum absolute value determines respective sets in each group being divided into according to data block
In the sub-step of the coding mode of pixel, if the maximum in absolute difference is less than or equal to 64, right
When group corresponding to this maximum is encoded, the complement of two's two's complement digit of the coding mode of employing is less than etc.
In 7, and when maximum is 0 or 1, the complement of two's two's complement digit of the coding mode of employing is 1.
According to the characteristic of reference frame, by the independent each data block calculating reference frame under different predictive modes
Compression ratio, the predictive mode that can select corresponding highest compression ratio individually compressed to data block, from
And effectively improve compression ratio and the compression success rate of reference frame.
Second embodiment of the invention is related to a kind of reference frame data compression method.In this embodiment,
Taking 8 × 8 data block as a example, specifically, comprise the following steps:
First, using one of candidate modes described in first embodiment, to except center pre-
Each pixel surveyed outside pixel calculates predictive value.
Second, the actual value of each pixel is deducted the difference that prediction is worth to each pixel.
3rd, according to current predictive mode, by 63 pixels in addition to the prediction pixel of center, draw
It is divided into a first kind group and 4 Equations of The Second Kind groups, wherein first kind group with Center Prediction pixel as boundary, draw
It is divided into two first kind subgroups, each Equations of The Second Kind group is divided into four Equations of The Second Kind subgroups by first kind group.
4th, calculate the coding mode of each subgroup that the 3rd step obtains.Specifically, it is determined that in subgroup
The scope of data of mm row in table 1 belonging to big absolute difference, corresponding to the scope of data of determination
Coding mode (code mode, cm) is the coding mode of this subgroup.Wherein, in Table 1, poor
Maximum in value absolute value is less than or equal to 64, then, when the group corresponding to this maximum being encoded, adopt
The complement of two's two's complement digit of coding mode is less than or equal to 7, and when maximum is 0 or 1, adopts
The complement of two's two's complement digit of coding mode is 1.S is sign bit.
Table 1. coding schedule
It is appreciated that in the other embodiment of the present invention, it would however also be possible to employ other coded systems are each
Subgroup determines coding mode.
5th, according to following manner, the coding mode of each subgroup is merged: first, by coding mode
Identical subgroup merges, and the unique subgroup of coding mode does not do merging treatment;Secondly, if the first kind
There is, in group and Equations of The Second Kind group, the subgroup that pixel number is 3, and there is a coding than this subgroup
The subgroup of pattern big 1, then by this two sub-portfolios simultaneously, and adopt larger coding mould for the group after merging
Formula.
6th, combination situation is obtained according to the 5th step, calculates each according to required for table 1 completes to encode
Bit number (table look-up by the coding mode, difference data and the table 1 that include all subgroups (including merging subgroup)
When necessary maximum symbol mark) and reform patterns coding required for bit number (first kind group is only
There are 2 kinds of reform patterns, need bit to represent, Equations of The Second Kind group has 15 kinds of reform patterns, Equations of The Second Kind combines
And if become a group being made up of two subgroups, use 3 bits, other situations 4 bits);
7th, after calculating other candidate modes compression data blocks by the way of the first to the 6th step
Bit number, selects compression ratio highest predictive mode as the final predictive mode of 8x8 data block.
8th, using selecting final predictive mode, reference frame is compressed, calculates the bit after compression
Number, including Center Prediction pixel value (8 bit), predictive mode (3 bit), 8x8 data block
Coding mode (15~54 ratios of reform patterns (13~17 bit), first kind group and all Equations of The Second Kind groups
Special), all differential codings and table look-up 1 when necessary maximum symbol mark, if bit number after compression
It is the motionless block of pressure more than 512 this blocks, with former data block as final compression result, otherwise according to upper
State sequentially by the data final compression result of setup action in order.
Additionally, adopting said method, 8x8 data block compression result is tested, test employs 8
Individual different 720p sequence, compared with existing compression method, compression ratio (cr) that the present invention obtains,
Compression exceedes half ratio (≤256) and the situation of compression Success Ratio (≤512) is as shown in table 2:
Table 2 compression result deck watch
Wherein, qp represents quantization parameter.From table 2 it can be seen that it is (a kind of existing with hac+sbt
Compression method) compare, with the increase of qp, mda+sfl (a kind of existing compression method) compresses
Rate (, divided by 512 gained, lower explanation compression ratio is higher for the bit number after data block compression) and compression are super
The advantage crossing half (the higher the better) has more and more substantially slowed down although compressing successful inferior position, but
It is to calculate according to the breadth of 720p, compress unsuccessfully number and still differ very many;With hac+sbt phase
The institute that the present invention maintains mda_sfl substantially is advantageous, compresses successful number and hac+sbt for ratio
It is sufficiently close to;Compared with mda+sfl, when qp is less than normal, the compression ratio of the present invention and compress successfully all
Advantageous, when qp is bigger than normal, although compression ratio is relatively poor, compresses successful advantage and still exist.
The each method embodiment of the present invention all can be realized in modes such as software, hardware, firmwares.No matter
The present invention is to be realized with software, hardware or firmware mode, and instruction code may be stored in any class
In the addressable memorizer of computer of type (for example permanent or revisable, volatibility or non-
Volatibility, solid-state or non-solid, fixing or removable medium etc.).Equally,
Memorizer may, for example, be programmable logic array (programmable array logic, referred to as
" pal "), random access memory (random access memory, referred to as " ram "),
Programmable read only memory (programmable read only memory, referred to as " prom "),
Read only memory (read-only memory, referred to as " rom "), electrically erasable are read-only
Memorizer (electrically erasable programmable rom, referred to as " eeprom "),
Disk, CD, digital versatile disc (digital versatile disc, referred to as " dvd ") etc..
Third embodiment of the invention is related to a kind of reference frame data compressor.Fig. 6 is this reference frame number
Structural representation according to compressor.
Specifically, this reference frame data compressor includes:
Cutting unit, for being divided into the data block of multiple predefined sizes by reference frame.
Compression unit, for each data block, when calculating the predictive mode applying various candidates respectively
Compression ratio, and select compression ratio highest predictive mode that this data block is compressed.Wherein,
The predictive mode of each candidate only carries out block interior prediction using each pixel in notebook data block, and each candidate's is pre-
Data block is divided into multiple groups by survey pattern respectively, and each group includes multiple pixels, the prediction of each candidate
Pattern includes in group pixel by the predictive mode of 45 degree or 135 degree directional spreding.
Additionally, in a preference, data block is organized interior pixel by 45 degree or 135 degree of directional spreding
Predictive mode segmentation multiple groups include:
Pixel edge in one group of first kind group being made up of the pixel on 135 degree of diagonal and at least two groups groups
45 degree of directional spreding are in the Equations of The Second Kind group of 135 degree of diagonal both sides, and the number of pixels in first kind group is
/ 2nd of number of pixels in Equations of The Second Kind group.Or one group is made up of the pixel on 45 degree of diagonal
Pixel in first kind group and at least two groups groups is along 135 degree of directional spreding in the of 45 degree of diagonal both sides
Two class groups, and the number of pixels in first kind group is 1/2nd of number of pixels in Equations of The Second Kind group.
In another preference, the predictive mode of each candidate includes the first rotation predictive mode, data block
Included by multiple groups of the first rotation predictive mode segmentation:
One group of first kind group and at least two being made up of the pixel in a line in this data block center two row
The Equations of The Second Kind group that group is made up of the adjacent two row pixels of this data block.Wherein, this predictive mode is initially pre-
Survey pixel and be located at the row that first kind group is located, first kind group is divided into two first kind by initial predicted pixel
Group, the prediction direction of two first kind subgroups is contrary, and the row that each Equations of The Second Kind group is located by first kind group divides
Become two Equations of The Second Kind subgroups, the prediction direction belonging to two Equations of The Second Kind subgroups of an Equations of The Second Kind group together is mutually vertical
Directly, the prediction direction and between subgroup is along changing clockwise or counterclockwise.
In another preference, the predictive mode of each candidate includes the second rotation predictive mode, data block
Included by multiple groups of the second rotation predictive mode segmentation:
One group arranged by this data block center two in string on the first kind group and at least two that forms of pixel
The Equations of The Second Kind group that group is made up of the adjacent rows pixel of this data block.Wherein, this predictive mode is initially pre-
Survey pixel and be located at the row that first kind group is located, first kind group is divided into two first kind by initial predicted pixel
Group, the prediction direction of two first kind subgroups is contrary, and the row that each Equations of The Second Kind group is located by first kind group divide
Become two Equations of The Second Kind subgroups, the prediction direction belonging to two Equations of The Second Kind subgroups of an Equations of The Second Kind group together is mutually vertical
Directly, the prediction direction and between subgroup is along changing clockwise or counterclockwise.
Additionally, in another preference, above-mentioned compression unit includes following subelement:
Computation subunit, for according to current prediction mode calculate data block in each pixel predictive value and
Difference between actual value.
Coded sub-units, the difference for maximum absolute value in each group of being divided into according to data block is true
Determine the coding mode of respective sets pixel.And coded sub-units are when determining coding mode, if difference is exhausted
64, then when the group corresponding to this maximum being encoded are less than or equal to the maximum in value, employing
The complement of two's two's complement digit of coding mode is less than or equal to 7, and when maximum is 0 or 1, employing
The complement of two's two's complement digit of coding mode is 1.
Compression subelement, carries out to each group encoding institute using the corresponding encoded pattern that each group determines for calculating
The bit number needing, and compression under current prediction mode for the data block is determined based on the bit number calculating gained
Rate.
Additionally, in another preference, data block includes nxn pixel, wherein n is more than 4
Positive integer.The initial predicted pixel of the predictive mode of each candidate is in the pixel at data block center one
Individual.
First embodiment is the method embodiment corresponding with present embodiment, and present embodiment can be with
First embodiment is worked in coordination enforcement.The relevant technical details mentioned in first embodiment are in this enforcement
Still effective in mode, in order to reduce repetition, repeat no more here.Correspondingly, carry in present embodiment
To relevant technical details be also applicable in first embodiment.
It should be noted that each unit mentioned in the present invention each equipment embodiment is all logical block,
Physically, a logical block can be the one of a physical location or a physical location
Part, can also be realized with the combination of multiple physical locations, these logical block physics realization sides of itself
Formula is not most important, and the combination of the function that these logical blocks are realized is only the solution present invention and is carried
The key of the technical problem going out.Additionally, for the innovative part projecting the present invention, the present invention is above-mentioned respectively to be set
The unit less close with solving technical problem relation proposed by the invention is not drawn by standby embodiment
Enter, this is not intended that the said equipment embodiment does not have other units.
It should be noted that in the claim and description of this patent, the first and second grades it
The relational terms of class are used merely to make a distinction an entity or operation with another entity or operation,
And not necessarily require or imply between these entities or operation, there is any this actual relation or suitable
Sequence.And, term " inclusion ", "comprising" or its any other variant are intended to nonexcludability
Comprise, so that including a series of process of key elements, method, article or equipment not only include that
A little key elements, but also include other key elements being not expressly set out, or also include for this process,
Method, article or the intrinsic key element of equipment.In the absence of more restrictions, by sentence " bag
Including one " key element that limits is being it is not excluded that including the process of described key element, method, article or setting
Also there is other identical element in standby.
Although by referring to some of the preferred embodiment of the invention, the present invention has been shown and
Description, but it will be understood by those skilled in the art that can in the form and details it be made respectively
Plant and change, without departing from the spirit and scope of the present invention.
Claims (12)
1. a kind of reference frame data compression method is it is characterised in that comprise the following steps:
Reference frame is divided into the data block of multiple predefined sizes;
To each data block, calculate the compression ratio during predictive mode applying various candidates respectively, and select
Select compression ratio highest predictive mode this data block is compressed;Wherein, the predictive mode of each candidate
Carry out block interior prediction using each pixel in notebook data block, the predictive mode of each candidate is respectively by data block
It is divided into multiple groups, each group includes multiple pixels, and the predictive mode of each candidate includes pixel in group
Press the predictive mode of 45 degree or 135 degree directional spreding.
2. reference frame data compression method according to claim 1 is it is characterised in that described number
Multiple groups split by described predictive mode according to block include:
Pixel in one group of first kind group being made up of the pixel on 135 degree of diagonal and at least two groups groups
Along 45 degree of directional spreding in the Equations of The Second Kind group of 135 degree of diagonal both sides, and the pixel count in first kind group
Mesh is 1/2nd of number of pixels in Equations of The Second Kind group;Or
Pixel edge in one group of first kind group being made up of the pixel on 45 degree of diagonal and at least two groups groups
135 degree of directional spreding are in the Equations of The Second Kind group of 45 degree of diagonal both sides, and the number of pixels in first kind group
For number of pixels in Equations of The Second Kind group 1/2nd.
3. reference frame data compression method according to claim 1 is it is characterised in that each candidate
Predictive mode include first rotation predictive mode, described data block by first rotation predictive mode split
Multiple groups include:
One group of first kind group and at least two being made up of the pixel in a line in this data block center two row
The Equations of The Second Kind group that group is made up of the adjacent two row pixels of this data block;Wherein, this predictive mode is initially pre-
Survey pixel and be located at the row that first kind group is located, first kind group is divided into two first kind by initial predicted pixel
Group, the prediction direction of described two first kind subgroups is contrary, and each Equations of The Second Kind group is located by first kind group
Row is divided into two Equations of The Second Kind subgroups, belongs to the prediction direction phase of two Equations of The Second Kind subgroups of an Equations of The Second Kind group together
Mutually vertical, and the prediction direction between described subgroup is along changing clockwise or counterclockwise.
4. reference frame data compression method according to claim 1 is it is characterised in that each candidate
Predictive mode include second rotation predictive mode, described data block by second rotation predictive mode split
Multiple groups include:
One group arranged by this data block center two in string on the first kind group and at least two that forms of pixel
The Equations of The Second Kind group that group is made up of the adjacent rows pixel of this data block;Wherein, this predictive mode is initially pre-
Survey pixel and be located at the row that first kind group is located, first kind group is divided into two first kind by initial predicted pixel
Group, the prediction direction of described two first kind subgroups is contrary, and each Equations of The Second Kind group is located by first kind group
Row are divided into two Equations of The Second Kind subgroups, belong to the prediction direction phase of two Equations of The Second Kind subgroups of an Equations of The Second Kind group together
Mutually vertical, and the prediction direction between described subgroup is along changing clockwise or counterclockwise.
5. reference frame data compression method according to claim 1 is it is characterised in that described "
To each data block, calculate the compression ratio during predictive mode applying various candidates respectively " step bag
Include following sub-step:
Difference between the predictive value of each pixel and actual value in data block is calculated according to current prediction mode
Value;
Respective sets picture is determined according to the described difference of maximum absolute value in each group that data block is divided into
The coding mode of element;
Calculate the bit number needed for each group being encoded using the corresponding encoded pattern that each group determines, and base
Determine compression ratio under current prediction mode for the data block in the described bit number calculating gained.
6. reference frame data compression method according to claim 5 is it is characterised in that described
Respective sets pixel is determined according to the described difference of maximum absolute value in each group that data block is divided into
In the sub-step of coding mode, if the maximum in absolute difference is less than or equal to 64, to this
When the corresponding group of big value is encoded, the complement of two's two's complement digit of the coding mode of employing is less than or equal to 7,
And when described maximum is 0 or 1, the complement of two's two's complement digit of the coding mode of employing is 1.
7. reference frame data compression method according to any one of claim 1 to 6, its feature
It is, described data block includes nxn pixel, wherein n is the positive integer more than 4;
The initial predicted pixel of the predictive mode of described each candidate is the pixel positioned at described data block center
One of.
8. a kind of reference frame data compressor is it is characterised in that include:
Cutting unit, for being divided into the data block of multiple predefined sizes by reference frame;
Compression unit, for each data block, when calculating the predictive mode applying various candidates respectively
Compression ratio, and select compression ratio highest predictive mode that this data block is compressed;
Wherein, the predictive mode of each candidate only carries out block interior prediction using each pixel in notebook data block,
Data block is divided into multiple groups by the predictive mode of each candidate respectively, and each group includes multiple pixels, respectively
The predictive mode of candidate includes in group pixel by the predictive mode of 45 degree or 135 degree directional spreding.
9. reference frame data compressor according to claim 8 is it is characterised in that each candidate
Predictive mode include first rotation predictive mode, described data block by first rotation predictive mode split
Multiple groups include:
One group of first kind group and at least two being made up of the pixel in a line in this data block center two row
The Equations of The Second Kind group that group is made up of the adjacent two row pixels of this data block;Wherein, this predictive mode is initially pre-
Survey pixel and be located at the row that first kind group is located, first kind group is divided into two first kind by initial predicted pixel
Group, the prediction direction of described two first kind subgroups is contrary, and each Equations of The Second Kind group is located by first kind group
Row is divided into two Equations of The Second Kind subgroups, belongs to the prediction direction phase of two Equations of The Second Kind subgroups of an Equations of The Second Kind group together
Mutually vertical, and the prediction direction between described subgroup is along changing clockwise or counterclockwise.
10. reference frame data compressor according to claim 8 is it is characterised in that each wait
The predictive mode of choosing includes the second rotation predictive mode, and described data block is divided by the second rotation predictive mode
Multiple groups cut include:
One group arranged by this data block center two in string on the first kind group and at least two that forms of pixel
The Equations of The Second Kind group that group is made up of the adjacent rows pixel of this data block;Wherein, this predictive mode is initially pre-
Survey pixel and be located at the row that first kind group is located, first kind group is divided into two first kind by initial predicted pixel
Group, the prediction direction of described two first kind subgroups is contrary, and each Equations of The Second Kind group is located by first kind group
Row are divided into two Equations of The Second Kind subgroups, belong to the prediction direction phase of two Equations of The Second Kind subgroups of an Equations of The Second Kind group together
Mutually vertical, and the prediction direction between described subgroup is along changing clockwise or counterclockwise.
11. reference frame data compressor according to claim 8 are it is characterised in that described
Compression unit includes following subelement:
Computation subunit, for according to current prediction mode calculate data block in each pixel predictive value and
Difference between actual value;
Coded sub-units, for the described difference of maximum absolute value in each group of being divided into according to data block
Value determines the coding mode of respective sets pixel;
Compression subelement, carries out to each group encoding institute using the corresponding encoded pattern that each group determines for calculating
The bit number needing, and determine data block under current prediction mode based on the described bit number calculating gained
Compression ratio;And
Described coded sub-units determine coding mode when, if the maximum in described absolute difference is little
In equal to 64, then, when the group corresponding to this maximum being encoded, the two of the coding mode of employing enters
Complement code digit processed is less than or equal to 7, and when described maximum is 0 or 1, the coding mode of employing
Complement of two's two's complement digit is 1.
The 12. reference frame data compressor any one of according to Claim 8 to 11, it is special
Levy and be, described data block includes nxn pixel, wherein n is the positive integer more than 4;
The initial predicted pixel of the predictive mode of described each candidate is the pixel positioned at described data block center
One of.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510435316.8A CN106375762B (en) | 2015-07-22 | 2015-07-22 | Reference frame data compression method and its device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510435316.8A CN106375762B (en) | 2015-07-22 | 2015-07-22 | Reference frame data compression method and its device |
Publications (2)
Publication Number | Publication Date |
---|---|
CN106375762A true CN106375762A (en) | 2017-02-01 |
CN106375762B CN106375762B (en) | 2019-05-24 |
Family
ID=57880155
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201510435316.8A Active CN106375762B (en) | 2015-07-22 | 2015-07-22 | Reference frame data compression method and its device |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106375762B (en) |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107582046A (en) * | 2017-09-18 | 2018-01-16 | 山东正心医疗科技有限公司 | ECG real-time monitor method |
CN108810556A (en) * | 2017-04-28 | 2018-11-13 | 炬芯(珠海)科技有限公司 | The method, apparatus and chip of compressed reference frame |
CN110007855A (en) * | 2019-02-28 | 2019-07-12 | 华中科技大学 | A kind of the 3D stacking NVM internal storage data compression method and system of hardware supported |
CN110175185A (en) * | 2019-04-17 | 2019-08-27 | 南京天数智芯科技有限公司 | A kind of self-adaptive non-loss compression based on time series data distribution characteristics |
CN110933444A (en) * | 2018-09-20 | 2020-03-27 | 北京君正集成电路股份有限公司 | Bit width value storage method and device |
CN110933413A (en) * | 2018-09-20 | 2020-03-27 | 北京君正集成电路股份有限公司 | Video frame processing method and device |
CN111866519A (en) * | 2019-04-30 | 2020-10-30 | 上海富瀚微电子股份有限公司 | Coding method |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101431671A (en) * | 2007-11-07 | 2009-05-13 | 财团法人工业技术研究院 | Methods for selecting a prediction mode and encoder thereof |
CN103125118A (en) * | 2010-10-01 | 2013-05-29 | 索尼公司 | Image processing device and image processing method |
CN103248891A (en) * | 2013-04-24 | 2013-08-14 | 复旦大学 | N-bit truncation quantization and in-block two-dimensional prediction-based reference frame compression method |
CN103563389A (en) * | 2011-06-03 | 2014-02-05 | 高通股份有限公司 | Intra prediction mode coding with directional partitions |
CN103988437A (en) * | 2011-11-08 | 2014-08-13 | 高通股份有限公司 | Context reduction for context adaptive binary arithmetic coding |
CN103997646A (en) * | 2014-05-13 | 2014-08-20 | 北京航空航天大学 | Rapid intra-frame prediction mode selection method in high-definition video coding |
-
2015
- 2015-07-22 CN CN201510435316.8A patent/CN106375762B/en active Active
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101431671A (en) * | 2007-11-07 | 2009-05-13 | 财团法人工业技术研究院 | Methods for selecting a prediction mode and encoder thereof |
CN103125118A (en) * | 2010-10-01 | 2013-05-29 | 索尼公司 | Image processing device and image processing method |
CN103563389A (en) * | 2011-06-03 | 2014-02-05 | 高通股份有限公司 | Intra prediction mode coding with directional partitions |
CN103988437A (en) * | 2011-11-08 | 2014-08-13 | 高通股份有限公司 | Context reduction for context adaptive binary arithmetic coding |
CN103248891A (en) * | 2013-04-24 | 2013-08-14 | 复旦大学 | N-bit truncation quantization and in-block two-dimensional prediction-based reference frame compression method |
CN103997646A (en) * | 2014-05-13 | 2014-08-20 | 北京航空航天大学 | Rapid intra-frame prediction mode selection method in high-definition video coding |
Cited By (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108810556A (en) * | 2017-04-28 | 2018-11-13 | 炬芯(珠海)科技有限公司 | The method, apparatus and chip of compressed reference frame |
CN108810556B (en) * | 2017-04-28 | 2021-12-24 | 炬芯科技股份有限公司 | Method, device and chip for compressing reference frame |
CN107582046A (en) * | 2017-09-18 | 2018-01-16 | 山东正心医疗科技有限公司 | ECG real-time monitor method |
CN110933444A (en) * | 2018-09-20 | 2020-03-27 | 北京君正集成电路股份有限公司 | Bit width value storage method and device |
CN110933413A (en) * | 2018-09-20 | 2020-03-27 | 北京君正集成电路股份有限公司 | Video frame processing method and device |
CN110933444B (en) * | 2018-09-20 | 2021-10-29 | 北京君正集成电路股份有限公司 | Bit width value storage method and device |
CN110007855A (en) * | 2019-02-28 | 2019-07-12 | 华中科技大学 | A kind of the 3D stacking NVM internal storage data compression method and system of hardware supported |
CN110007855B (en) * | 2019-02-28 | 2020-04-28 | 华中科技大学 | Hardware-supported 3D stacked NVM (non-volatile memory) memory data compression method and system |
CN110175185A (en) * | 2019-04-17 | 2019-08-27 | 南京天数智芯科技有限公司 | A kind of self-adaptive non-loss compression based on time series data distribution characteristics |
CN110175185B (en) * | 2019-04-17 | 2023-04-07 | 上海天数智芯半导体有限公司 | Self-adaptive lossless compression method based on time sequence data distribution characteristics |
CN111866519A (en) * | 2019-04-30 | 2020-10-30 | 上海富瀚微电子股份有限公司 | Coding method |
Also Published As
Publication number | Publication date |
---|---|
CN106375762B (en) | 2019-05-24 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106375762A (en) | Reference frame data compression method and apparatus | |
CN111800641B (en) | Image coding and decoding method and device adopting different kinds of reconstructed pixels in same mode | |
CN103797800B (en) | It derives reference model value and coding and decoding represents the information of prediction mode | |
WO2015032350A1 (en) | Image compression method and device using block-matching | |
CN106412579B (en) | A kind of coding of image, coding/decoding method and device | |
JP2006246431A (en) | Image coding apparatus and method | |
CN108712648B (en) | Rapid intra-frame coding method for depth video | |
KR101946598B1 (en) | Image coding and decoding method and device | |
CN113784124B (en) | Block matching encoding and decoding method for fine division using multi-shape sub-blocks | |
CN101877792B (en) | Intra mode prediction method and device and coder | |
CN103067715A (en) | Encoding and decoding methods and encoding and decoding device of range image | |
CN102611880A (en) | Encoding method and device for marking geometric classification mode of image block | |
JP5652101B2 (en) | Image processing apparatus and image processing method | |
CN102271251B (en) | Lossless image compression method | |
CN111432212A (en) | Intra-frame division method and system based on texture features and storage medium | |
CN102542528B (en) | Image conversion processing method and system | |
CN114079792A (en) | Compression method, coding method and device for video coding reference block | |
CN117014618A (en) | Image compression-based blocking method and system and electronic equipment | |
CN107682699A (en) | A kind of nearly Lossless Image Compression method | |
US8515187B2 (en) | Method, compressor, decompressor and signal representation for lossless compression of pixel block values using row and column slope codewords | |
CN107666612A (en) | Block content categorizing method based on convolutional neural networks | |
CN104809747B (en) | The statistical method and its system of image histogram | |
CN111010574B (en) | Image compression method and device and electronic equipment | |
CN104683812B (en) | Video preprocessing method and device for motion estimation | |
CN113691811A (en) | Coding block dividing method, device, system and storage medium |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | 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 |