CN111954000A - Lossless compression method for high-speed toll collection picture set - Google Patents
Lossless compression method for high-speed toll collection picture set Download PDFInfo
- Publication number
- CN111954000A CN111954000A CN202010646265.4A CN202010646265A CN111954000A CN 111954000 A CN111954000 A CN 111954000A CN 202010646265 A CN202010646265 A CN 202010646265A CN 111954000 A CN111954000 A CN 111954000A
- Authority
- CN
- China
- Prior art keywords
- picture
- block
- data
- img
- original
- 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
Images
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
- H04N19/70—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals characterised by syntax aspects related to video coding, e.g. related to compression standards
-
- G—PHYSICS
- G07—CHECKING-DEVICES
- G07B—TICKET-ISSUING APPARATUS; FARE-REGISTERING APPARATUS; FRANKING APPARATUS
- G07B15/00—Arrangements or apparatus for collecting fares, tolls or entrance fees at one or more control points
- G07B15/06—Arrangements for road pricing or congestion charging of vehicles or vehicle users, e.g. automatic toll systems
-
- 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
- H04N19/17—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 an image region, e.g. an object
- H04N19/176—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 an image region, e.g. an object the region being a block, e.g. a macroblock
Abstract
The invention discloses a lossless compression method for a high-speed toll picture set, which compresses and stores redundant data through similarity calculation twice and picture blocking. According to the characteristics that camera equipment of the same toll lane is fixed and the backgrounds of pictures are highly similar, a toll bitmap in the same toll lane within a certain time period is constructed into a set, first similarity calculation is carried out, mainly pairwise similarity between pictures is calculated, a reference picture with the highest average similarity with other pictures is searched, then all pictures including the reference picture are partitioned, secondary similarity calculation is carried out on the picture partitions through comparison of data of each picture partition and the reference picture partitions, identical blocks are not stored repeatedly, and difference lossless compression storage is carried out on the highly similar blocks in a centralized mode. The invention can realize lossless compression of the picture and simultaneously give consideration to the saving property and the economical efficiency of storage.
Description
Technical Field
The invention relates to the field of image processing, in particular to a lossless compression method for a high-speed toll collection picture set.
Background
The bitmap is stored in a lossless manner by generally adopting RGB 3 bytes to represent a pixel point; for picture matrixes with much the same value, lossless compression by adopting run-length coding to reduce redundancy is the mainstream technology.
The use of bitmap format to store high-speed toll pictures respectively results in a large amount of identical information being stored repeatedly, which is a huge drain on storage resources.
For high-speed charged pictures, a large number of lossless bitmaps consume a lot of storage, and where there is a high similarity of content and information between different pictures, duplicate storage is made in physical storage.
Disclosure of Invention
The technical problem to be solved by the present invention is to provide a lossless compression method for a high-speed toll collection picture set, aiming at the defects of the prior art, the lossless compression method for the high-speed toll collection picture set can find redundant and repeated information, perform lossless compression on the picture set, and reduce the consumption of storage resources.
In order to solve the technical problems, the invention adopts the technical scheme that:
a lossless compression method for a high-speed toll collection picture set comprises the following steps.
Step (ii) of1. Establishing an original picture set: in the set time period of each toll lane, N automobiles pass through, each automobile is shot with one original picture, and then the same camera obtains N original pictures which are respectively IMG1,IMG2,…,IMGN. The resolution of each original picture is W multiplied by L, and the original pictures are RGB bitmaps. Establishing an original picture set IMG for the N pictures, wherein the IMG is equal to { IMG }1,IMG2,…,IMGN}。
Step 2, finding a reference picture, comprising the following steps:
step 21, calculating the similarity for the first time: and carrying out similarity calculation on any two original pictures in the original picture set IMG to obtain a similarity coefficient between the two original pictures.
Step 22, constructing a similarity matrix S: for all the similarity coefficients obtained in step 21, an N × N similarity matrix S is constructed. The N column vectors of each row in the similarity matrix S respectively represent the similarity coefficients of a certain original picture and N original pictures.
Step 23, finding a reference picture: respectively calculating the average value of each row in the similarity matrix S, selecting the original picture of the row with the maximum average value as a reference picture, and recording as XBaseline. Rearranging N-1 pictures except the reference picture in the original picture set IMG, and establishing a new picture set X, wherein X is { X ═ X1,X2,…,XN-1}。
Step 3, picture blocking: using a square window, the X formed in step 23BaselineAnd each picture in the new picture set X is divided into M blocks of data and numbered sequentially from top to bottom and from left to right. XBaselineRecording the picture after being blocked as X'BaselineIf the new picture set X is a blocked picture set denoted as X ', X ' ═ X '1,X′2,…,X′N-1}。
Step 4, calculating the secondary similarity: mixing M block data blocks of each picture in the block picture set X 'with X'BaselineComparing the M block data blocks one by one, and calculating each block data block and X 'of each picture in X'BaselineAnd the similarity coefficient sita of each block of data.
Step 5, creating a centralized storage data block and a data index matrix, specifically as follows:
step 5A, creating a data block CBaselineFor storing X'BaselineThe block data of (2).
Step 5B, creating a data block COrigFor storing X'iThe original tile data.
Step 5C, creating a data block CΔFor storing X 'and X'BaselineThe difference block data.
Step 5D, creating an index matrix STYPE (st) with dimensions of (N-1) multiplied by Mij) And the storage mode is used for storing each block in X'.
Step 5E, creating an index matrix IDX (ix) with (N-1) multiplied by M dimensionsij) For storing the storage location of each block in X'.
Step 6, redundancy removal storage: and storing the jth block data of the ith picture in the block picture set X' according to the similarity coefficient sita calculated in the step 4 as follows. Wherein i is more than or equal to 1 and less than or equal to N-1. J is more than or equal to 1 and less than or equal to M.
And step 61, when sita is equal to 100%, the jth block data block in the ith picture is not stored. Index matrix STYPE (st)ij) Storage mode value st inijIndex matrix IDX (ix) noted 0ij) Of (2) a storage location value ixijAnd recording as the corresponding sequence number j of the own data block.
Step 62, when the sita is less than or equal to gamma, storing the original data of the jth data block in the ith picture in the data block C created in the step 5OrigIn (1). Wherein gamma is a set similarity coefficient threshold value>50 percent. Index matrix STYPE (st)ij) Storage mode value st inijIndex matrix IDX (ix) noted as-1ij) Of (2) a storage location value ixijEquals data block COrigThe corresponding sequential block number in (1).
Step 63, when gamma is less than sita and less than 100%, pixel points of the jth block data block in the ith picture are firstly compared with X'BaselineMiddle j block data blockThe pixel points of (1) are subtracted one by one to form a jth pixel difference block. Then, the jth block of pixel difference values is stored in the data block C created in step 5ΔIn (1). Index matrix STYPE (st)ij) Storage mode value st inijIndex matrix IDX (ix) as 1ij) Of (2) a storage location value ixijIs marked as CΔAnd the sequence block sequence number corresponding to the data block.
Step 7, data block CΔLossless compression: after the N-1 pictures in the block picture set X' are subjected to redundancy removal storage according to the method in the step 6, run length coding is adopted to carry out redundancy storage on the data block CΔLossless compression is carried out, and the compressed result is marked as CΔ-code。
In step 21, during the similarity calculation for one time, the calculation formula of the similarity coefficient between the two original pictures is as follows:
wherein s isijRepresenting IMGiAnd IMGjThe similarity coefficient of (2). IMGiAnd IMGjAny two original pictures in the original picture set IMG. IMGi-IMGjAnd (4) subtracting the pixel points corresponding to the i and j original pictures one by one to form a pixel point difference value. count [ (IMG)i-IMGj)＝＝0]Indicating the number of pixel point differences of 0. W × L represents the resolution of each original picture in the original picture set IMG.
In step 22, the constructed similarity matrix S is expressed as follows:
S＝{sij}i,j∈(1，2，…，N)。
in step 23, the calculation formula for respectively calculating the average value for each row in the similarity matrix S is as follows:
wherein mean _ SiDenotes the i-th row in the similarity matrix S divided by SiiThe outer average value.
The calculation formula of the row i where the row average value in the similarity matrix S is the maximum is as follows:
iBaseline＝arg(Maxi(mean_S)),i∈(1,2,…,N)
wherein iBaselineIndicates a picture number corresponding to the reference picture.
In step 4, the calculation formula of the similarity coefficient sita is as follows:
wherein the content of the first and second substances,
j-th block data block and X 'representing ith picture in block picture set X'BaselineAnd the difference value of the pixel point of the jth block of data. R × R represents the resolution of each of the M block data blocks.
In step 62, the set similarity coefficient threshold gamma is 70% to 75%.
Further comprising step 8, Picture XiThe decompression method specifically comprises the following steps:
step 81, CΔ-codeDecompressing: from compressed data block CΔ-c；deDecompressing to obtain a data block CΔ。
Step 82, reading index information: reading the index matrix STYPE (st)ij) And an index matrix IDX (ix)ij) The storage mode information and the storage position information of the ith row.
Step 83, extracting block data: sequentially extracting data block CBaseline、COrig、CΔCorresponding block data.
Step 84, CΔAnd restoring the intermediate difference value block data: for C obtained in step 83ΔDifference block of
Obtaining a tile picture X 'by inverse operation'iOriginal tile data BLK in (B)i。
Step 85, X'iReduction: according to the read index information, the data block C is divided intoOrigAnd original tile data BLKiReduced to X'iData block CBaselineReduced to X'Baseline。
Step 86, binding X'iAnd X'BaselineRestoring the original picture Xi。
The invention has the following beneficial effects: and performing compressed storage on the redundant data through similarity calculation and picture blocking twice. According to the characteristics that camera equipment of the same toll lane is fixed and the backgrounds of pictures are highly similar, a toll bitmap in the same toll lane within a certain time period is constructed into a set, first similarity calculation is carried out, mainly pairwise similarity between pictures is calculated, a reference picture with the highest average similarity with other pictures is searched, then all pictures including the reference picture are partitioned, secondary similarity calculation is carried out on the picture partitions through comparison of data of each picture partition and the reference picture partitions, identical blocks are not stored repeatedly, and difference lossless compression storage is carried out on the highly similar blocks in a centralized mode. The invention can realize lossless compression of the picture and simultaneously give consideration to the saving property and the economical efficiency of storage.
Drawings
Fig. 1 shows a flow chart of a high-speed toll-like picture set oriented lossless compression method of the present invention.
Detailed Description
The present invention will be described in further detail with reference to the accompanying drawings and specific preferred embodiments.
In the description of the present invention, it is to be understood that the terms "left side", "right side", "upper part", "lower part", etc., indicate orientations or positional relationships based on those shown in the drawings, and are only for convenience of describing the present invention and simplifying the description, but do not indicate or imply that the referred device or element must have a specific orientation, be constructed in a specific orientation, and be operated, and that "first", "second", etc., do not represent an important degree of the component parts, and thus are not to be construed as limiting the present invention. The specific dimensions used in the present example are only for illustrating the technical solution and do not limit the scope of protection of the present invention.
As shown in fig. 1, a high-speed toll-free picture set oriented lossless compression method includes the following steps.
Step 1, establishing an original picture set: in a set time period (which can be 15 minutes, 30 minutes, 1 hour and the like) of each toll lane, N automobiles pass through the toll lane, and each automobile is shot by one original picture, so that N original pictures, namely IMG (inertial measurement group) are obtained by shooting by the same camera1,IMG2,…,IMGN. The resolution of each original picture is W multiplied by L, the original pictures are stored in an RGB bitmap without loss, and 24 bits in a storage medium represent 1 pixel value.
Establishing an original picture set IMG for the N pictures, wherein the IMG is equal to { IMG }1,IMG2,…,IMGN}。
Step 2, finding a reference picture, comprising the following steps:
step 21, calculating the similarity for the first time: and carrying out similarity calculation on any two original pictures in the original picture set IMG to obtain a similarity coefficient between the two original pictures.
In the first similarity calculation, the calculation formula of the similarity coefficient between the two original pictures is preferably as follows:
wherein s isijRepresenting IMGiAnd IMGjThe similarity coefficient of (2). IMGiAnd IMGjAny two original pictures in the original picture set IMG. IMGi-IMGjAnd (4) subtracting the pixel points corresponding to the i and j original pictures one by one to form a pixel point difference value. count [ (IMG)i-IMGj)＝＝0]Indicating the number of pixel point differences of 0. W × L represents the resolution of each original picture in the original picture set IMG.
Step 22, constructing a similarity matrix S: for all the similarity coefficients obtained in step 21, an N × N similarity matrix S is constructed. The N column vectors of each row in the similarity matrix S respectively represent the similarity coefficients of a certain original picture and N original pictures.
The similarity matrix S constructed as described above is preferably expressed as follows:
S＝{sij}i,j∈(1，2，…，N)。
wherein, the first row in the similarity matrix S represents the similarity coefficient of the first original picture and the N original pictures, which are respectively S11、s12、s13、……、s1N(ii) a Wherein s is11And 1, and so on.
And step 23, searching a reference picture.
Respectively calculating the average value of each row in the similarity matrix S, selecting the original picture of the row with the maximum average value as a reference picture, and recording as XBaseline。
The above calculation formula for respectively averaging each row in the similarity matrix S is preferably:
wherein mean _ SiDenotes the i-th row in the similarity matrix S divided by SiiThe outer average; as in the first row, divide s is calculated11In addition, the rest s12、s13、……、s1NAverage value of (a).
The calculation formula of the row i where the row average value in the similarity matrix S is the maximum is as follows:
iBaseline＝arg(Maxi(mean_S)),i∈(1,2,…,N)
wherein iBaselineIndicates a picture number corresponding to the reference picture.
Then, rearranging the N-1 pictures except the reference picture in the original picture set IMG, and creating a new picture set X, where X is ═ X1,X2,…,XN-1}。
And step 3, picture blocking.
Because each picture isThe pixels are arranged according to resolution ratio, so that the pixels are partitioned by a square window. The invention adopts non-overlapping blocks, the blocks are sequentially divided from top to bottom and from left to right, and the edge of the picture is not enough to complement 0 of the size of the block window. Usually T.times.R (R.epsilon.Z) is used+) The square window of (2) is usually a combination of (4 × 4,8 × 8,16 × 16) and the like in the resolution of the view picture.
According to this principle, let X be assumedBaselineThe matrixes after X blocking are respectively X'Baseline，X′:{X′1,X′2,…,X′N-1Dividing each picture into M blocks, numbering the M blocks in sequence, and sequencing the M blocks from top to bottom and from left to right.
And 4, calculating the secondary similarity.
Mixing M block data blocks of each picture in the block picture set X 'with X'BaselineComparing the M block data blocks one by one, and calculating each block data block and X 'of each picture in X'BaselineAnd the similarity coefficient sita of each block of data.
The formula for calculating the similarity coefficient sita is preferably:
wherein the content of the first and second substances,
j-th block data block and X 'representing ith picture in block picture set X'BaselineAnd the difference value of the pixel point of the jth block of data. R × R represents the resolution of each of the M block data blocks.
Step 5, creating a centralized storage data block and a data index matrix, specifically as follows:
step 5A, creating a data block CBaselineFor storing X'BaselineThe block data of (2). Wherein BLK is usedBaseline(j) Denotes reference picture X'BaselineWhere j ∈ M.
Step 5B, creating a data block COrigFor storing X'iThe original tile data.
Step 5C, creating a data block CΔFor storing X 'and X'BaselineThe difference block data. Wherein the content of the first and second substances,
is X'iAnd X'BaselineJ-th block of the difference block of (1,2, …, N-1), where i e (1,2, …, M). BLKi(j) Is X'iJ-th block of the original data block, where i ∈ (1,2, …, N-1), j ∈ (1,2, …, M).
Step 5D, creating an index matrix STYPE (st) with dimensions of (N-1) multiplied by Mij) And the storage mode is used for storing each block in X'.
Step 5E, creating an index matrix IDX (ix) with (N-1) multiplied by M dimensionsij) For storing the storage location of each block in X'.
Step 6, redundancy removal storage: and storing the jth block data of the ith picture in the block picture set X' according to the similarity coefficient sita calculated in the step 4 as follows. Wherein i is more than or equal to 1 and less than or equal to N-1. J is more than or equal to 1 and less than or equal to M.
And step 61, when sita is equal to 100%, the jth block data block in the ith picture is not stored. Index matrix STYPE (st)ij) Storage mode value st inijIndex matrix IDX (ix) noted 0ij) Of (2) a storage location value ixijAnd recording as the corresponding sequence number j of the own data block.
Step 62, when the sita is less than or equal to gamma, storing the original data of the jth data block in the ith picture in the data block C created in the step 5OrigIn (1). Wherein, gamma is a set similarity coefficient threshold, and the specific value of gamma can be determined according to the classification of the sample data setThe value of the analysis result is usually above 50%, preferably 70-75%.
Index matrix STYPE (st)ij) Storage mode value st inijIndex matrix IDX (ix) noted as-1ij) Of (2) a storage location value ixijEquals data block COrigThe corresponding sequential block number in (1).
Step 63, when gamma is less than sita and less than 100%, pixel points of the jth block data block in the ith picture are firstly compared with X'BaselineAnd subtracting the pixel points of the jth block of data one by one to form a jth pixel difference block. Then, the jth block of pixel difference values is stored in the data block C created in step 5ΔIn (1). Index matrix STYPE (st)ij) Storage mode value st inijIndex matrix IDX (ix) as 1ij) Of (2) a storage location value ixijIs marked as CΔAnd the sequence block sequence number corresponding to the data block.
Repeating the step 6 until the redundancy removal storage of the N-1 pictures in the block picture set X' is completed, wherein the specific calculation program preferably represents as follows:
for i in(1,2,…,N-1):
Δ′i＝X′i-X′Baseline
for j in(1,2,…,M):
if sita＝100％：
stij＝0
ixij＝j
if sita>gamma&sita<100％:
stij＝1
ixij＝order1
order1＝order1+1
else:
stij＝-1
ixij＝order2
CΔ(order1)<＝BLKi(j)
ordEr2＝order2+1
wherein, the constants ordEr1 and ordEr2 respectively represent CΔAnd COrigThe initial values of the block numbers of (1) are 0, respectively.
Step 7, data block CΔAnd (4) lossless compression.
CΔIs a matrix containing a large number of continuous 0 value elements, and after N-1 pictures in the block picture set X' are subjected to redundancy removal and storage according to the method in the step 6, the data block C is subjected to run length codingΔLossless compression is carried out, and the compressed result is marked as CΔ-code。
The run-length coding principle and process are as follows:
assuming that 9BYTE data of {00,00,00,00, 5C,9A } data represents 3 pixel point difference values, and the data which can be changed into 6BYTE data of {07,00,01,5C,01,9A } data after run length coding completes the representation of the pixel point difference values, the storage space is reduced by more than 30%, wherein {07,00} represents 7 00BYTE values, and {01,5C } represents 1 5C BYTE value, and so on.
After calculation by the above cycle, CBaseline、COrig、CΔ-code、STYPE(stij)、IDX(ixij) The relevant information is stored.
Step 8, Picture XiThe decompression method specifically comprises the following steps:
step 81, CΔ-c；deDecompressing: from compressed data block CΔ-c；deDecompressing to obtain a data block CΔ。
Step 82, reading index information: reading the index matrix STYPE (st)ij) And an index matrix IDX (ix)ij) The storage mode information and the storage position information of the ith row.
Step 83, extracting block data: sequentially extracting data block CBaseline、COrig、CΔCorresponding block data.
Step 84, CΔAnd restoring the intermediate difference value block data: for C obtained in step 83ΔDifference block of
Obtaining a tile picture X 'by inverse operation'iOriginal tile data BLK in (B)i。
Step 85, X'iReduction: according to the read index information, the data block C is divided intoOrigAnd original tile data BLKiReduced to X'iData block CBaselineReduced to X'Baseline。
Step 86, binding X'iAnd X'BaselineAnd according to the resolution, restoring the original picture Xi。
According to the algorithm, through simulation tests, when R takes a value of 16 and gamma takes a value of 75%, the storage space of bitmap storage can be reduced by 12% at most in the same hour of the same high-speed toll lane under the premise of lossless compression.
Although the preferred embodiments of the present invention have been described in detail, the present invention is not limited to the details of the embodiments, and various equivalent modifications can be made within the technical spirit of the present invention, and the scope of the present invention is also within the scope of the present invention.
Claims (7)
1. A lossless compression method for high-speed toll collection picture sets is characterized in that: the method comprises the following steps:
step 1, establishing an original picture set: in the set time period of each toll lane, N automobiles pass through, each automobile is shot with one original picture, and then the same camera obtains N original pictures which are respectively IMG1，IMG2，…，IMGN(ii) a The resolution of each original picture is W multiplied by L, and the original pictures are RGB bitmaps; establishing an original picture set IMG for the N pictures, wherein the IMG is equal to { IMG }1，IMG2，…，IMGN}；
Step 2, finding a reference picture, comprising the following steps:
step 21, calculating the similarity for the first time: carrying out similarity calculation on any two original pictures in the original picture set IMG to obtain a similarity coefficient between the two original pictures;
step 22, constructing a similarity matrix S: constructing an NxN similarity matrix S aiming at all the similarity coefficients obtained in the step 21; n column vectors of each row in the similarity matrix S respectively represent similarity coefficients of a certain original picture and N original pictures;
step 23, finding a reference picture: respectively calculating the average value of each row in the similarity matrix S, selecting the original picture of the row with the maximum average value as a reference picture, and recording as XBaseline(ii) a Rearranging N-1 pictures except the reference picture in the original picture set IMG, and establishing a new picture set X, wherein X is { X ═ X1，X2，…，XN-1}；
Step 3, picture blocking: using a square window, the X formed in step 23BaselineDividing each picture in the new picture set X into M blocks of data and numbering the M blocks of data in sequence from top to bottom and from left to right; xBaselineRecording the picture after being blocked as X'BaselineIf the new picture set X is a blocked picture set denoted as X ', X ' ═ X '1，X′2，…，X′N-1}；
Step 4, calculating the secondary similarity: mixing M block data blocks of each picture in the block picture set X 'with X'BaselineComparing the M block data blocks one by one, and calculating each block data block and X 'of each picture in X'BaselineThe similarity coefficient sita of each block of data;
step 5, creating a centralized storage data block and a data index matrix, specifically as follows:
step 5A, creating a data block CBaselineFor storing X'BaselineThe block data of (2);
step 5B, creating a data block COrigFor storing X'iThe original tile data;
step 5C,Creating a data Block CΔFor storing X 'and X'BaselineThe difference block data;
step 5D, creating an index matrix STYPE (st) with dimensions of (N-1) multiplied by Mij) A storage means for storing each block in X';
step 5E, creating an index matrix IDX (ix) with (N-1) multiplied by M dimensionsij) A storage location for storing each block in X';
step 6, redundancy removal storage: storing the jth block data block of the ith picture in the block picture set X' according to the similarity coefficient sita calculated in the step 4 in the following mode; wherein i is more than or equal to 1 and less than or equal to N-1; j is more than or equal to 1 and less than or equal to M;
step 61, when sita is equal to 100%, the jth block data block in the ith picture is not stored; index matrix STYPE (st)ij) Storage mode value st inijIndex matrix IDX (ix) noted 0ij) Of (2) a storage location value ixijRecording as the corresponding sequence number j of the data block;
step 62, when the sita is less than or equal to gamma, storing the original data of the jth data block in the ith picture in the data block C created in the step 5OrigPerforming the following steps; wherein gamma is a set similarity coefficient threshold value, and is more than 50 percent; index matrix STYPE (st)ij) Storage mode value st inijIndex matrix IDX (ix) as 1ij) Of (2) a storage location value ixijEquals data block COrigThe sequence number of the corresponding sequential block in (1);
step 63, when gamma is less than sita and less than 100%, pixel points of the jth block data block in the ith picture are firstly compared with X'BaselineSubtracting the pixel points of the jth block of data one by one to form a jth pixel difference block; then, the jth block of pixel difference values is stored in the data block C created in step 5ΔPerforming the following steps; index matrix STYPE (st)ij) Storage mode value st inijIndex matrix IDX (ix) as 1ij) Of (2) a storage location value ixijIs marked as CΔSequence block numbers corresponding to the data blocks;
step 7, data block CΔLossless compression: when N-1 pictures in the block picture set X' are all processed according to the method of the step 6After the redundant storage is finished, the data block C is coded by run lengthΔLossless compression is carried out, and the compressed result is marked as CΔ-code。
2. The high-speed charged picture set oriented lossless compression method as claimed in claim 1, wherein: in step 21, during the similarity calculation for one time, the calculation formula of the similarity coefficient between the two original pictures is as follows:
wherein s isijRepresenting IMGiAnd IMGjThe similarity coefficient of (2); IMGiAnd IMGjAny two original pictures in the original picture set IMG are taken; IMGi-IMGjSubtracting pixels corresponding to the i and j original pictures one by one to form pixel difference values; count [ (IMG)i-IMGj)＝＝0]Representing the number of pixel point difference values of 0; w × L represents the resolution of each original picture in the original picture set IMG.
3. The high-speed toll-oriented picture set lossless compression method as claimed in claim 2, wherein: in step 22, the constructed similarity matrix S is expressed as follows:
S＝{sij}i，j∈(1，2，…，N)。
4. the high-speed toll-oriented picture set lossless compression method as claimed in claim 3, wherein: in step 23, the calculation formula for respectively calculating the average value for each row in the similarity matrix S is as follows:
wherein mean _ SiDenotes the i-th row in the similarity matrix S divided by SiiThe outer average;
the calculation formula of the row i where the row average value in the similarity matrix S is the maximum is as follows:
iBaseline＝arg(Maxi(mean_S))，i∈(1，2，…，N)
wherein iBaselineIndicates a picture number corresponding to the reference picture.
5. The high-speed charged picture set oriented lossless compression method as claimed in claim 1, wherein: in step 4, the calculation formula of the similarity coefficient sita is as follows:
wherein the content of the first and second substances,
j-th block data block and X 'representing ith picture in block picture set X'BaselineThe difference value of the pixel point of the middle jth data block; r × R represents the resolution of each of the M block data blocks.
6. The high-speed charged picture set oriented lossless compression method as claimed in claim 1, wherein: in step 62, the set similarity coefficient threshold gamma is 70% to 75%.
7. The high-speed charged picture set oriented lossless compression method as claimed in claim 1, wherein: further comprising step 8, Picture XiThe decompression method specifically comprises the following steps:
step 81, CΔ-codeDecompressing: from compressed data block CΔ-codeDecompressing to obtain a data block CΔ；
Step 82, reading index information: reading the index matrix STYPE (st)ij) And an index matrix IDX (ix)ij) The storage mode information and the storage position information of the ith row;
step 83, extract the number of blocksAccording to the following steps: sequentially extracting data block CBaseline、COrig、CΔCorresponding block data;
step 84, CΔAnd restoring the intermediate difference value block data: for C obtained in step 83ΔDifference block of
Obtaining a tile picture X 'by inverse operation'iOriginal tile data BLK in (B)i；
Step 85, X'iReduction: according to the read index information, the data block C is divided intoOrigAnd original tile data BLKiReduced to X'iData block CBaselineReduced to X'Baseline；
Step 86, binding X'iAnd X'BaselineRestoring the original picture Xi。
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010646265.4A CN111954000B (en) | 2020-07-07 | 2020-07-07 | Lossless compression method for high-speed toll collection picture set |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010646265.4A CN111954000B (en) | 2020-07-07 | 2020-07-07 | Lossless compression method for high-speed toll collection picture set |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111954000A true CN111954000A (en) | 2020-11-17 |
| CN111954000B CN111954000B (en) | 2021-04-27 |
Family
ID=73340189
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010646265.4A Active CN111954000B (en) | 2020-07-07 | 2020-07-07 | Lossless compression method for high-speed toll collection picture set |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111954000B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113420183A (en) * | 2021-06-21 | 2021-09-21 | 成都神州数码索贝科技有限公司 | Cross-video target retrieval and collision method and system in video detection system |
Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6263105B1 (en) * | 1996-10-04 | 2001-07-17 | Sony Corporation | Image processing apparatus and method |
| CN1710960A (en) * | 2005-07-21 | 2005-12-21 | 中华电信股份有限公司 | Continuous image compressing method with difference judgement |
| CN1763749A (en) * | 2004-10-20 | 2006-04-26 | 微软公司 | Analyzing operational data and another data of a search system |
| CN101594537A (en) * | 2009-06-04 | 2009-12-02 | 京北方科技股份有限公司 | Massive image data compression method |
| CN102831222A (en) * | 2012-08-24 | 2012-12-19 | 华中科技大学 | Differential compression method based on data de-duplication |
| CN102842123A (en) * | 2012-07-12 | 2012-12-26 | 南京理工大学 | Sparse-region residual error compensating and revising method for improving marginal definition during image sampling |
| CN103810694A (en) * | 2012-11-15 | 2014-05-21 | 腾讯科技（深圳）有限公司 | Quality factor obtaining method and device in image compression |
| CN104796726A (en) * | 2015-04-10 | 2015-07-22 | 北京邮电大学 | Method for detecting image compression traces |
| CN104869425A (en) * | 2015-05-13 | 2015-08-26 | 信阳师范学院 | Compression and decompression method based on texture image similarity |
| CN105407353A (en) * | 2014-09-11 | 2016-03-16 | 腾讯科技（深圳）有限公司 | Image compression method and apparatus |
| CN108875828A (en) * | 2018-06-19 | 2018-11-23 | 太原学院 | A kind of fast matching method and system of similar image |
| CN110087076A (en) * | 2019-04-09 | 2019-08-02 | 华南理工大学 | A kind of more method for compressing image based on hierarchical block replacement |
-
2020
- 2020-07-07 CN CN202010646265.4A patent/CN111954000B/en active Active
Patent Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6263105B1 (en) * | 1996-10-04 | 2001-07-17 | Sony Corporation | Image processing apparatus and method |
| CN1763749A (en) * | 2004-10-20 | 2006-04-26 | 微软公司 | Analyzing operational data and another data of a search system |
| CN1710960A (en) * | 2005-07-21 | 2005-12-21 | 中华电信股份有限公司 | Continuous image compressing method with difference judgement |
| CN101594537A (en) * | 2009-06-04 | 2009-12-02 | 京北方科技股份有限公司 | Massive image data compression method |
| CN102842123A (en) * | 2012-07-12 | 2012-12-26 | 南京理工大学 | Sparse-region residual error compensating and revising method for improving marginal definition during image sampling |
| CN102831222A (en) * | 2012-08-24 | 2012-12-19 | 华中科技大学 | Differential compression method based on data de-duplication |
| CN103810694A (en) * | 2012-11-15 | 2014-05-21 | 腾讯科技（深圳）有限公司 | Quality factor obtaining method and device in image compression |
| CN105407353A (en) * | 2014-09-11 | 2016-03-16 | 腾讯科技（深圳）有限公司 | Image compression method and apparatus |
| CN104796726A (en) * | 2015-04-10 | 2015-07-22 | 北京邮电大学 | Method for detecting image compression traces |
| CN104869425A (en) * | 2015-05-13 | 2015-08-26 | 信阳师范学院 | Compression and decompression method based on texture image similarity |
| CN108875828A (en) * | 2018-06-19 | 2018-11-23 | 太原学院 | A kind of fast matching method and system of similar image |
| CN110087076A (en) * | 2019-04-09 | 2019-08-02 | 华南理工大学 | A kind of more method for compressing image based on hierarchical block replacement |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113420183A (en) * | 2021-06-21 | 2021-09-21 | 成都神州数码索贝科技有限公司 | Cross-video target retrieval and collision method and system in video detection system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111954000B (en) | 2021-04-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US6438268B1 (en) | Vector quantization codebook generation method | |
| JP4486780B2 (en) | Nonlinear quantization and similarity matching method for image information retrieval | |
| US11221990B2 (en) | Ultra-high compression of images based on deep learning | |
| CN110111256B (en) | Image super-resolution reconstruction method based on residual distillation network | |
| EP1559273A4 (en) | Non-linear quantization and similarity matching methods for retrieving video sequence having a set of image frames | |
| CN108491430B (en) | Unsupervised Hash retrieval method based on clustering characteristic directions | |
| JP2003523587A (en) | Visual attention system | |
| GB2300083A (en) | Fractal image compression | |
| CN104160409A (en) | Method and system for image analysis | |
| US5574573A (en) | Compression method for a standardized image library | |
| CN110162290B (en) | Compression method for DeMURA data of OLED screen | |
| CN114745553A (en) | Image data storage method based on big data | |
| CN111954000B (en) | Lossless compression method for high-speed toll collection picture set | |
| CN112132279A (en) | Convolutional neural network model compression method, device, equipment and storage medium | |
| CN109523016B (en) | Multi-valued quantization depth neural network compression method and system for embedded system | |
| CN110728354A (en) | Improved sliding type grouping convolution neural network | |
| Geman et al. | Invariant statistics and coding of natural microimages | |
| CN109583584B (en) | Method and system for enabling CNN with full connection layer to accept indefinite shape input | |
| George et al. | Speeding-up Fractal Color Image Compression Using Moments Features Based on Symmetry Predictor | |
| KR20040015911A (en) | A Blind Watermarking Method by Grouping Codewords for VQ-Quantized Images | |
| Battiato et al. | A bio-inspired CNN with re-indexing engine for lossless DNA microarray compression and segmentation | |
| CN111986280B (en) | Image compression method based on maximum inter-class variance block compressed sensing | |
| JPH08167028A (en) | Image processing method | |
| CN110780838B (en) | Self-adaptive DeMURA data compression method for OLED screen | |
| CN116758618B (en) | Image recognition method, training device, electronic equipment and storage medium |
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 |