CN110691242A - Large-format remote sensing image lossless compression method - Google Patents

Large-format remote sensing image lossless compression method Download PDF

Info

Publication number
CN110691242A
CN110691242A CN201910871670.3A CN201910871670A CN110691242A CN 110691242 A CN110691242 A CN 110691242A CN 201910871670 A CN201910871670 A CN 201910871670A CN 110691242 A CN110691242 A CN 110691242A
Authority
CN
China
Prior art keywords
value
independent coding
pixel
coding region
lossless compression
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
Application number
CN201910871670.3A
Other languages
Chinese (zh)
Other versions
CN110691242B (en
Inventor
颜露新
陈立群
张天序
钟胜
李旭
崔裕宾
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Huazhong University of Science and Technology
Original Assignee
Huazhong University of Science and Technology
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Huazhong University of Science and Technology filed Critical Huazhong University of Science and Technology
Priority to CN201910871670.3A priority Critical patent/CN110691242B/en
Publication of CN110691242A publication Critical patent/CN110691242A/en
Application granted granted Critical
Publication of CN110691242B publication Critical patent/CN110691242B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

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
    • H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
    • H04N19/102—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or selection affected or controlled by the adaptive coding
    • H04N19/124—Quantisation
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
    • H04N19/134—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or criterion affecting or controlling the adaptive coding
    • H04N19/136—Incoming video signal characteristics or properties
    • H04N19/137—Motion inside a coding unit, e.g. average field, frame or block difference
    • 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
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
    • H04N19/169—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding
    • H04N19/182—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being a pixel
    • 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
    • H04N19/184—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 bits, e.g. of the compressed video stream
    • 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
    • H04N19/90—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using coding techniques not provided for in groups H04N19/10-H04N19/85, e.g. fractals
    • H04N19/98—Adaptive-dynamic-range coding [ADRC]

Landscapes

  • Engineering & Computer Science (AREA)
  • Multimedia (AREA)
  • Signal Processing (AREA)
  • Compression Or Coding Systems Of Tv Signals (AREA)

Abstract

The invention discloses a large-format remote sensing image lossless compression method, and belongs to the field of image processing. The method comprises the following steps: dividing the large-format remote sensing image into independent coding regions which are not overlapped with each other; identifying the dynamic range of each independent coding region pixel, and determining the pixel quantization bit number; calculating lossless compression parameters of the independent coding regions according to the quantization bit number of each independent coding region; and carrying out lossless compression on the independent coding regions according to the lossless compression parameters of each independent coding region until the lossless compression of the large-format remote sensing image is completed. The large-format remote sensing image is divided into a plurality of independent coding regions which are not overlapped with each other, the dynamic range of each independent coding region is identified and is used as the local dynamic range of the large-format remote sensing image, the characteristic that the local dynamic range of the large-format remote sensing image is lower than the fixed dynamic range of a camera is effectively utilized, the overall lossless compression ratio is improved, the capability of adjusting compression parameters according to the local dynamic range is realized, and the lossless compression of the large-format remote sensing image is realized.

Description

Large-format remote sensing image lossless compression method
Technical Field
The invention belongs to the field of image processing, and particularly relates to a large-format remote sensing image lossless compression method.
Background
With the continuous development of China in the space fields of earth observation, deep space exploration and the like, the resolution of remote sensing images is increasingly improved, and the data volume is increasingly increased. The contradiction between the huge data volume of the large-format remote sensing image and the satellite-ground limited download bandwidth becomes a main contradiction restricting the development of the space field, and the image compression technology is an effective way for solving the problem.
The lossless compression ratio is closely related to the dynamic range of the camera image. In general, for the same scene, the smaller the number of camera image quantization bits, the higher the lossless compression ratio; the larger the number of quantization bits of the camera image, the lower the lossless compression ratio. Compression parameters of traditional lossless compression methods (e.g., JPEG-LS algorithm, context adaptive lossless image compression CALIC algorithm) are based on the number of quantization bits N of the camera imagebAnd (5) fixing and setting. For example, for 16bit camera image data, lossless compression parameters are set according to a dynamic range of 0-65535. In fact, in the large-format remote sensing image, the local dynamic range of the deep space background and the night area is obviously lower than the global dynamic range of the camera image; especially for remote sensing images in midnight time period, the dynamic range of the whole area is obviously lower than the fixed dynamic range of the camera. For these low dynamic range image local regions, the conventional lossless compression method still uses the phaseThe parameters are set in the fixed dynamic range of the machine for compression, and lossless compression ratio is low. If the compression parameters are adaptively adjusted according to the local dynamic range of the image, the advantage of small local range of the image can be effectively utilized, and the overall lossless compression ratio is improved.
Patent CN105828070A proposes a JPEG-LS block compression method, which divides the image into non-overlapping sub-blocks, and independently performs lossless or near lossless compression, so as to achieve lossless or near lossless compression of local areas of the image, improve the flexibility of compression, and prevent error code diffusion. However, each image sub-block is still compressed in a uniform and fixed dynamic range, and the ability of adjusting compression parameters according to a local dynamic range is not provided, so that the compression ratio is low.
Disclosure of Invention
Aiming at the defects and the improvement requirements of the prior art, the invention provides a large-format remote sensing image lossless compression method, aiming at solving the problem that each image sub-block is compressed by adopting a uniform and fixed dynamic range, so that the compression ratio is low.
To achieve the above object, according to one aspect of the present invention, there is provided a large format remote sensing image lossless compression method, including the steps of:
s1, dividing a large-format remote sensing image into independent coding regions which are not overlapped with each other;
s2, identifying the dynamic range of each independent coding region pixel, and determining the pixel quantization bit number;
s3, calculating lossless compression parameters of the independent coding regions according to the quantization bit number of each independent coding region;
and S4, performing lossless compression on the independent coding regions according to the lossless compression parameters of each independent coding region until the lossless compression of the large-format remote sensing image is completed.
Specifically, step S2 specifically includes:
and carrying out OR calculation on all pixel values in the current independent coding region according to bit positions to obtain the highest bit with the bit position of 1 in the calculation result, wherein the bit sequence is increased by 1 from 0, namely the minimum pixel bit number bpp of the current independent coding region.
Specifically, step S3 includes the following sub-steps:
s31, setting the micro-damage degree Near to 0;
s32, calculating the maximum value MaxVal which is possible to appear in the pixel value of the current independent coding region according to the minimum pixel bit number bpp of the current independent coding region, wherein the calculation formula is as follows:
MaxVal＝2bpp-1
s33, calculating a residual modulus parameter Range according to the micro-loss Near and the maximum value MaxVal, wherein the calculation formula is as follows:
s34, calculating a Golomb coding length limit parameter according to the minimum pixel bit number bpp and the residual modular parameter Range of the current independent coding region, wherein the calculation formula is as follows:
Limit＝2×(bpp+max(8，bpp))
Figure BDA0002203011340000032
s35, calculating the initial value of the context parameter according to the modulus parameter Range of the residual error, wherein the calculation formula is as follows:
Figure BDA0002203011340000033
B[Q]＝C[Q]＝0
N[Q]＝1
s36, calculating an address quantization threshold value T according to the maximum value MaxVal1、T2、T3The calculation formula is as follows:
T1＝FACTOR+2
T2＝FACTOR×4+3
T3＝FACTOR×17+4
Figure BDA0002203011340000034
wherein, Limit is the maximum value of the code length of the single-pixel Golomb code, qbpp is the quantization bit number of the pixel residual error, A [ Q ], B [ Q ], C [ Q ] and N [ Q ] respectively store the sum of the absolute values of the prediction residual errors with the context address of Q, the sum of the prediction residual errors, the prediction correction value and the occurrence frequency of the context, and FACTOR is the multiple FACTOR of the address quantization threshold value.
Specifically, step S4 includes the following sub-steps:
s41, calculating to obtain a residual mapping value ME and a Golomb coding parameter k of the independent coding region according to the lossless compression parameter of the independent coding region;
s42, carrying out Golomb length-limited coding on each pixel of the independent coding region according to the residual mapping value ME and the Golomb coding parameter k;
s43, sequentially traversing to the next independent coding region, and repeating the steps S41-S42 until the lossless compression of all independent coding regions of the large-format remote sensing image is completed.
Specifically, step S41 includes the following sub-steps:
(1) selecting four pixels R adjacent to the current pixel in the directions of west, north, northwest and northeasta、Rb、Rc、RdCalculating a gradient value D1＝Rd-Rb、D2＝Rb-Rc、D3＝Rc-Ra；
(2) According to each DiAnd each address quantization threshold value TiCalculating a context address value Q and a context symbol Sign, i is 1,2, 3;
(3) according to Ra、Rb、RcSolving a fixed predicted value Px by using a median predictor;
(4) obtaining a prediction correction value P' x ═ min ((Px + Sign × C [ Q ]), MaxVal) according to the prediction correction value C [ Q ], the context Sign and the fixed prediction value Px;
(5) calculating a prediction residual error E which is equal to Sign x (Ix-P 'x) according to the P' x, the pixel true value Ix of the current pixel and the context Sign Sign;
(6) according to the micro-damage Near and the prediction residual E, a residual quantization value E _ q is calculated
Figure BDA0002203011340000041
(7) Calculating a reconstruction value Rx according to the micro-loss Near, the residual quantized value E _ q and the context Sign, wherein the calculation process is as follows:
Rx＝Px+Sign*E_q*(2*Near+1)
if Rx is less than 0, Rx is 0; if Rx > MaxVal, then Rx is maxVal;
(8) performing modulus operation on the residual error E _ q according to the residual error modulus parameter Range to obtain a residual error modulus value E _ mod;
(9) according to the context parameter A [ Q ]]、N[Q]Computing Golomb coding parameters
Figure BDA0002203011340000042
(10) Mapping the modulus value E _ mod of the residual error to be a non-negative value to obtain a residual error mapping value ME;
(11) and updating the context parameters A [ Q ], B [ Q ], C [ Q ] and N [ Q ] according to the residual modulus value E _ Q of the current pixel.
Specifically, the step (2) is as follows:
if D isi≤-T3Then Q isi＝-4；
If D isi≤-T2Then Q isi＝-3；
If D isi≤-T1Then Q isi＝-2；
If D isiIf < -Near, then Qi＝-1；
If D isiLess than or equal to Near, then Qi＝0；
If D isi＜T1Then Q isi＝1；
If D isi＜T2Then Q isi＝2；
If D isi＜T3Then Q isi＝3；
else，Qi＝4。
Specifically, the step (10) is as follows:
if ((Near ═ 0) and (k ≦ -nq)) are satisfied and (E _ mod ≧ 0), ME ≦ 2 × E _ mod + 1;
when ((Near ═ 0) and (k ═ 0) are satisfied, and (2 × B [ Q ≦ -nq ])) and (E _ mod ≧ 0) are not satisfied, ME × (E _ mod + 1);
if ((Near ═ 0) and (k ≦ 0) and (2 × B [ Q ≦ -nq ])) are not satisfied and (E _ mod ≧ 0), ME is 2 × E _ mod;
if ((Near ═ 0) and (k ≦ 0) and (2 × B [ Q ≦ -nq ])) are not satisfied and (E _ mod ≧ 0), ME is 2 × E _ mod-1.
Specifically, the step (11) is as follows:
B[Q]＝B[Q]+E_q×(2×Near+1)；
A[Q]＝A[Q]+abs(E_q)；
if N [ Q ] ═ RESET is satisfied, then A [ Q ] ═ A [ Q ] > 1;
if N [ Q ] ═ RESET is satisfied and B [ Q ] ≧ 0 is satisfied, then B [ Q ] ═ B [ Q ] > 1;
if N [ Q ] ═ RESET is satisfied and B [ Q ] ≧ 0 is not satisfied, then B [ Q ] ═ - ((1-B [ Q ]) > 1), N [ Q ] > 1;
N[Q]＝N[Q]+1；
if B [ Q ] is less than or equal to-N [ Q ], then B [ Q ] + N [ Q ];
if B [ Q ] ≦ -NQ and C [ Q ] > -128 are satisfied, then C [ Q ] ═ CQ ] -1;
if B [ Q ] is not more than-NQ ] and B [ Q ] is not more than-NQ ], then B [ Q ] is-NQ + 1;
if B [ Q ] > 0 is satisfied, then B [ Q ] ═ B [ Q ] -N [ Q ];
if B [ Q ] is equal to or less than-N [ Q ] and C [ Q ] is less than 127, then C [ Q ] + 1;
if B [ Q ] is less than or equal to-N [ Q ] and B [ Q ] is more than 0, then B [ Q ] is equal to 0;
where RESET is a JPEG-LS encoding parameter, and > denotes a right shift operator.
Specifically, step S42 includes the following sub-steps:
s421, calculating Golomb coding parameters val and n according to the residual mapping value ME and the Golomb coding parameter k, wherein the calculation formula is as follows:
Figure BDA0002203011340000061
n＝ME-val×2k
s422, carrying out Golomb length-limited coding on the E according to val and n to obtain a code stream;
and S423, outputting the minimum pixel bit number bpp and the first pixel FirstPixel of the current independent coding region together as side information and the code stream of the current independent coding region.
To achieve the above object, according to another aspect of the present invention, there is provided a computer-readable storage medium having stored thereon a computer program which, when executed by a processor, implements the large format remote sensing image lossless compression method according to the first aspect.
Generally, by the above technical solution conceived by the present invention, the following beneficial effects can be obtained:
(1) the invention divides the large-format remote sensing image into a plurality of independent coding regions which do not overlap with each other, because the code streams of different independent coding regions are independent from each other, when the code streams have errors in transmission, the decoding image of the independent coding region corresponding to the position of the error code stream is only influenced, and the error code is limited in one independent coding region, thereby effectively increasing the error code resistance.
(2) The dynamic range of each independent coding region is identified, the dynamic range of the independent coding region is used as the local dynamic range of the large-format remote sensing image, the characteristic that the local dynamic range of the large-format remote sensing image is lower than the fixed dynamic range of a camera is effectively utilized, the overall lossless compression ratio is improved, the method has the capability of adjusting compression parameters according to the local dynamic range, and the lossless compression of the large-format remote sensing image is realized.
(3) The invention carries out or calculates all pixel values in the independent coding region according to the bit positions, can adaptively identify the dynamic range of the local region of the large-format remote sensing image, sets the lossless compression parameter according to the dynamic range, further utilizes the characteristic that the local dynamic range of the large-format remote sensing image is lower than the fixed dynamic range of the camera, and improves the integral lossless compression ratio.
(4) The invention takes the minimum pixel bit number of the independent coding region as the side information, and considers the dynamic range of the local region of the large-format remote sensing image, so that the side information and the code stream can be combined for decoding subsequently.
Drawings
Fig. 1 is a flowchart of a large-format remote sensing image lossless compression method according to an embodiment of the present invention;
FIG. 2 is a schematic diagram illustrating selection of four neighboring values of a current pixel according to an embodiment of the present invention;
fig. 3 is a schematic diagram of a large-format remote sensing image according to an embodiment of the present invention;
FIG. 4 is a minimum number of pixel bits bpp histogram for an independently encoded region according to an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention. In addition, the technical features involved in the embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other.
As shown in fig. 1, the present invention provides a large-format remote sensing image lossless compression method, which includes the following steps:
and S1, dividing the large-format remote sensing image into independent coding regions which are not overlapped with each other.
Dividing a large-format remote sensing image Img with the pixel size of R multiplied by C into independent coding regions ICR which are not overlapped and have the pixel size of R multiplied by CiP, where P is the total number of independent encoding regions, R, C is the row and column values of image Img, respectively, and r, c are the independent encoding regions ICR, respectivelyiThe row and column values of (1) are set as required.
The specific implementation process is as follows: dividing the image Img into a plurality of independent coding regions which do not overlap with each other according to the set row value R (R is more than 0 and less than R) and the set column value C (C is more than 0 and less than C) of the independent coding regions. When the row value R of the image Img cannot be divided by the row value R of the independent coding region, filling the last row of pixel values of the current independent coding region; when the image Img column value C cannot be divided exactly by the independently encoded section column value C, the last column of pixel values of the current independently encoded section is used for padding.
And S2, identifying the dynamic range of each independent coding region pixel, and determining the pixel quantization bit number.
Obtaining current independent coding region image ICRiStatistical independent coding region ICRiDynamic range of pixel, obtaining ICR capable of representing current independent coding regioniThe minimum number of pixel bits bpp for all pixels.
The specific implementation process is as follows: and carrying out OR calculation on all pixel values in the current independent coding region according to bit positions to obtain the highest bit with the bit position of 1 in the calculation result, wherein the bit sequence (starting from 0) plus 1 is the minimum pixel bit number bpp value of the current coding region. For example, assuming that all pixel values of the current coding region are bitwise ored into 126, the highest bit of the 126 bits is 6, and the minimum pixel bit number bpp of the current coding region is 6+1 ═ 7.
And S3, calculating lossless compression parameters of the independent coding regions according to the quantization bit number of each independent coding region.
Calculating the ICR of the current independent coding region according to the minimum pixel bit number bpp of the independent coding region obtained by identificationiThe lossless compression parameter of (1).
Step S3 includes the following substeps:
and S31, setting the micro-damage degree Near to 0.
The micro-impairment Near is set to 0, indicating lossless compression.
S32, calculating the maximum value MaxVal which is possible to appear in the pixel value of the current independent coding region according to the minimum pixel bit number bpp of the current independent coding region, wherein the calculation formula is as follows:
MaxVal＝2bpp-1
s33, calculating a residual modulus parameter Range according to the micro-loss Near and the maximum value MaxVal, wherein the calculation formula is as follows:
s34, calculating a Golomb coding length limit parameter according to the minimum pixel bit number bpp and the residual modular parameter Range of the current independent coding region, wherein the calculation formula is as follows:
Limit＝2×(bpp+max(8，bpp))
Figure BDA0002203011340000092
wherein, Limit is the maximum value of the code length of the single-pixel Golomb code, and qbpp is the quantization bit number of the pixel residual error.
S35, calculating the initial value of the context parameter according to the modulus parameter Range of the residual error, wherein the calculation formula is as follows:
Figure BDA0002203011340000093
B[Q]＝C[Q]＝0
N[Q]＝1
wherein, A [ Q ], B [ Q ], C [ Q ] and N [ Q ] respectively store the sum of absolute values of prediction residuals with context address of Q, the sum of the prediction residuals, the prediction correction value and the number of times of the occurrence of the context, and are updated in the following steps.
S36, calculating an address quantization threshold value T according to the maximum value MaxVal1、T2、T3。
T1＝FACTOR+2
T2＝FACTOR×4+3
T3＝FACTOR×17+4
Wherein, FACTOR is a multiple FACTOR of the address quantization threshold, and the calculation formula is as follows:
Figure BDA0002203011340000101
and acquiring the dynamic range of the sub-image, adaptively determining lossless compression parameters, and performing lossless compression on the sub-image. And performing the operation on each block sub-image to complete the whole image compression. The lossless compression parameter setting is based on the dynamic range of the image, and a better compression effect can be obtained by adopting a smaller dynamic range.
And S4, performing lossless compression on the independent coding regions according to the lossless compression parameters of each independent coding region until the lossless compression of the large-format remote sensing image is completed.
And carrying out lossless compression on the independent coding region according to the calculated lossless compression parameters to obtain a code stream, and combining and outputting the code stream with information such as the first pixel FirstPixel and the minimum pixel bit number bpp of the current independent coding region. The first pixel FirstPixel is the first pixel of the independently encoded region.
Step S4 includes the following substeps:
and S41, calculating to obtain a residual mapping value ME and a Golomb coding parameter k of the independent coding region according to the lossless compression parameter of the independent coding region.
(1) Selecting four pixels R adjacent to the current pixel in the directions of west, north, northwest and northeasta、Rb、Rc、RdCalculating a gradient value D1、D2、D3。
As shown in FIG. 2, four adjacent values R of the current pixel are selected according to the JPEG-LS context templatea、Rb、Rc、RdCalculating three gradient values D1、D2、D3。
D1＝Rd-Rb
D2＝Rb-Rc
D3＝Rc-Ra
(2) According to each DiAnd each address quantization threshold value TiThe context address value Q and the context Sign, i are calculated to be 1,2, 3.
According to DiAnd address quantization threshold T1、T2、T3Calculating an address quantization value QiI is 1,2,3, which is as follows:
if D isi≤-T3Then Q isi-4; if D isi≤-T2Then Q isi-3; if D isi≤-T1Then Q isi-2; if D isiIf < -Near, then Qi-1; if D isiLess than or equal to Near, then Q i0; if D isi＜T1Then Q isi1 is ═ 1; if D isi＜T2Then Q isi2; if D isi＜T3Then Q isi＝3；else，Qi＝4。
Quantizing the value Q according to the addressiThe context address value Q and the context Sign are computed.
Q＝|81×Q1+9×Q2+Q3|
Figure BDA0002203011340000111
(3) According to Ra、Rb、RcAnd solving a fixed predicted value Px by using a median predictor.
Figure BDA0002203011340000112
(4) Based on the prediction correction value C [ Q ], the context Sign, and the fixed prediction value Px, a prediction correction value P' x ═ min ((Px + Sign × C [ Q ]), MaxVal) is obtained.
(5) And calculating a prediction residual error E which is equal to Sign x (Ix-P 'x) according to the P' x, the pixel true value Ix of the current pixel and the context Sign Sign.
(6) According to the micro-damage Near and the prediction residual E, a residual quantization value E _ q is calculated
Figure BDA0002203011340000113
(7) Calculating a reconstruction value Rx according to the micro-loss Near, the residual quantized value E _ q and the context Sign, wherein the calculation process is as follows:
Rx＝Px+Sign*E_q*(2*Near+1)
if Rx is less than 0, Rx is 0; if Rx > MaxVal, then Rx is maxVal;
(8) performing modulus operation on the residual error E _ q according to the residual error modulus parameter Range to obtain a residual error modulus value E _ mod;
Figure BDA0002203011340000114
(9) according to the context parameter A [ Q ]]、N[Q]Computing Golomb coding parameters
(10) And mapping the modulus value E _ mod of the residual error into a non-negative value to obtain a residual error mapping value ME.
If (Near ═ 0) and (k ═ 0) are satisfied and (2 × B [ Q ≦ -nq ]) and (E _ mod ≧ 0), ME is 2 × E _ mod + 1;
when (Near ═ 0) and (k ═ 0) are satisfied, and (2 × B [ Q ≦ -nq ]) and (E _ mod ≧ 0), ME is 2 × (E _ mod + 1);
if ((Near ═ 0) and (k ≦ 0) and (2 × B [ Q ≦ -nq ])) are not satisfied and (E _ mod ≧ 0), ME is 2 × E _ mod;
if ((Near ═ 0) and (k ≦ 0) and (2 × B [ Q ≦ -nq ])) are not satisfied and (E _ mod ≧ 0), ME is 2 × E _ mod-1.
(11) And updating the context parameters A [ Q ], B [ Q ], C [ Q ] and N [ Q ] according to the residual modulus value E _ Q of the current pixel.
B[Q]＝B[Q]+E_q×(2×Near+1)；
A[Q]＝A[Q]+abs(E_q)；
If N [ Q ] ═ RESET is satisfied, then A [ Q ] ═ A [ Q ] > 1;
if N [ Q ] ═ RESET is satisfied and B [ Q ] ≧ 0 is satisfied, then B [ Q ] ═ B [ Q ] > 1;
if N [ Q ] ═ RESET is satisfied and B [ Q ] ≧ 0 is not satisfied, then B [ Q ] ═ - ((1-B [ Q ]) > 1), N [ Q ] > 1;
N[Q]＝N[Q]+1；
if B [ Q ] is less than or equal to-N [ Q ], then B [ Q ] + N [ Q ];
if B [ Q ] ≦ -NQ and C [ Q ] > -128 are satisfied, then C [ Q ] ═ CQ ] -1;
if B [ Q ] is not more than-NQ ] and B [ Q ] is not more than-NQ ], then B [ Q ] is-NQ + 1;
if B [ Q ] > 0 is satisfied, then B [ Q ] ═ B [ Q ] -N [ Q ];
if B [ Q ] is equal to or less than-N [ Q ] and C [ Q ] is less than 127, then C [ Q ] + 1;
if B [ Q ] is less than or equal to-N [ Q ] and B [ Q ] is more than 0, then B [ Q ] is equal to 0;
where the value of the JPEG-LS encoding parameter RESET is default or set as desired, the > denotes the right shift operator.
And S42, carrying out Golomb length-limited coding on each pixel of the independent coding region according to the residual mapping value ME and the Golomb coding parameter k.
S421, calculating Golomb coding parameters val and n according to the residual mapping value ME and the Golomb coding parameter k, wherein the calculation formula is as follows:
Figure BDA0002203011340000131
n＝ME-val×2k
and carrying out Golomb length-limited coding, wherein the length is limited to Limit-qbpp-1.
S422, carrying out Golomb length-limited coding on the E according to val and n to obtain a code stream;
and S423, outputting the minimum pixel bit number bpp and the first pixel FirstPixel of the current independent coding region together as side information and the code stream of the current independent coding region.
And performing lossless compression on the independent coding region to obtain a code stream, and combining the minimum pixel quantization bit number of the independent coding region as side information with the code stream.
S43, sequentially traversing to the next independent coding region, and repeating the steps S41-S42 until the lossless compression of all independent coding regions of the large-format remote sensing image is completed.
And finishing lossless compression of the large-format remote sensing image to obtain the combination of the side information and the code stream of the large-format remote sensing image.
Correspondingly, the invention provides a large-format remote sensing image lossless decompression method, which comprises the following steps:
and decoding (the decoding is decompression, the problem is compression, and no specific decompression is involved) by using the side information and code stream combination of the large-format remote sensing image to obtain the original data of the large-format remote sensing image.
Fig. 3 shows a large format remote sensing image with resolution 1354 × 2030 and a pixel bit width of 16 bits (fig. 3 in table 1). Let the row value r of the independent coding region be 16, and the column value c be 64, count the minimum pixel bit number bpp histogram of the independent coding region for the image (the camera pixel bit width is 16 bits), and the result is shown in fig. 4, where 0.3% is 12 bits, 62.1% is 13 bits, 37.4% is 14 bits, and only 0.2% is 16 bits. The minimum number of pixel bits bpp for most of the independently encoded regions is less than the camera bit width 16.
The patent CN105828070A and the invention are used for carrying out lossless compression on the large-format remote sensing image, the pixel row value of the independent coding region is set to be 16, the column value is set to be 64, the compression ratio comparison result is shown in Table 1, and the result shows that the lossless compression ratio of the invention is larger than the lossless compression ratio of the patent CN 105828070A.
Figure BDA0002203011340000141
TABLE 1
And the lossless compression parameters are adaptively adjusted according to the local dynamic range, so that the lossless compression ratio of the whole image is improved. Experiments show that compared with the CN105828070A patent which adopts fixed and unchangeable lossless compression parameters, the compression ratio is improved by about 10 percent.
It will be understood by those skilled in the art that the foregoing is only a preferred embodiment of the present invention, and is not intended to limit the invention, and that any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the scope of the present invention.

Claims (10)

1. A large-format remote sensing image lossless compression method is characterized by comprising the following steps:
s1, dividing a large-format remote sensing image into independent coding regions which are not overlapped with each other;
s2, identifying the dynamic range of each independent coding region pixel, and determining the pixel quantization bit number;
s3, calculating lossless compression parameters of the independent coding regions according to the quantization bit number of each independent coding region;
and S4, performing lossless compression on the independent coding regions according to the lossless compression parameters of each independent coding region until the lossless compression of the large-format remote sensing image is completed.
2. The method according to claim 1, wherein step S2 is specifically:
and carrying out OR calculation on all pixel values in the current independent coding region according to bit positions to obtain the highest bit with the bit position of 1 in the calculation result, wherein the bit sequence is increased by 1 from 0, namely the minimum pixel bit number bpp of the current independent coding region.
3. The method according to claim 1 or 2, characterized in that step S3 comprises the sub-steps of:
s31, setting the micro-damage degree Near to 0;
s32, calculating the maximum value MaxVal which is possible to appear in the pixel value of the current independent coding region according to the minimum pixel bit number bpp of the current independent coding region, wherein the calculation formula is as follows:
MaxVal＝2bpp-1
s33, calculating a residual modulus parameter Range according to the micro-loss Near and the maximum value MaxVal, wherein the calculation formula is as follows:
Figure FDA0002203011330000011
s34, calculating a Golomb coding length limit parameter according to the minimum pixel bit number bpp and the residual modular parameter Range of the current independent coding region, wherein the calculation formula is as follows:
Limit＝2×(bpp+max(8,bpp))
Figure FDA0002203011330000012
s35, calculating the initial value of the context parameter according to the modulus parameter Range of the residual error, wherein the calculation formula is as follows:
Figure FDA0002203011330000021
B[Q]＝C[Q]＝0
N[Q]＝1
s36, calculating an address quantization threshold value T according to the maximum value MaxVal1、T2、T3The calculation formula is as follows:
T1＝FACTOR+2
T2＝FACTOR×4+3
T3＝FACTOR×17+4
wherein, Limit is the maximum value of the code length of the single-pixel Golomb code, qbpp is the quantization bit number of the pixel residual error, A [ Q ], B [ Q ], C [ Q ] and N [ Q ] respectively store the sum of the absolute values of the prediction residual errors with the context address of Q, the sum of the prediction residual errors, the prediction correction value and the occurrence frequency of the context, and FACTOR is the multiple FACTOR of the address quantization threshold value.
4. The method of claim 3, wherein step S4 includes the sub-steps of:
s41, calculating to obtain a residual mapping value ME and a Golomb coding parameter k of the independent coding region according to the lossless compression parameter of the independent coding region;
s42, carrying out Golomb length-limited coding on each pixel of the independent coding region according to the residual mapping value ME and the Golomb coding parameter k;
s43, sequentially traversing to the next independent coding region, and repeating the steps S41-S42 until the lossless compression of all independent coding regions of the large-format remote sensing image is completed.
5. The method of claim 4, wherein step S41 includes the sub-steps of:
(1) selecting four pixels R adjacent to the current pixel in the directions of west, north, northwest and northeasta、Rb、Rc、RdCalculating a gradient value D1＝Rd-Rb、D2＝Rb-Rc、D3＝Rc-Ra；
(2) According to each DiAnd each address quantization threshold value TiCalculating a context address value Q and a context symbol Sign, i is 1,2, 3;
(3) according to Ra、Rb、RcSolving a fixed predicted value Px by using a median predictor;
(4) obtaining a prediction correction value P' x ═ min ((Px + Sign × C [ Q ]), MaxVal) according to the prediction correction value C [ Q ], the context Sign and the fixed prediction value Px;
(5) calculating a prediction residual error E which is equal to Sign x (Ix-P 'x) according to the P' x, the pixel true value Ix of the current pixel and the context Sign Sign;
(6) according to the micro-damage Near and the prediction residual E, a residual quantization value E _ q is calculated
Figure FDA0002203011330000031
(7) Calculating a reconstruction value Rx according to the micro-loss Near, the residual quantized value E _ q and the context Sign, wherein the calculation process is as follows:
Rx＝Px+Sign*E_q*(2*Near+1)
if Rx is less than 0, Rx is 0; if Rx > MaxVal, then Rx is maxVal;
(8) performing modulus operation on the residual error E _ q according to the residual error modulus parameter Range to obtain a residual error modulus value E _ mod;
(9) according to the context parameter A [ Q ]]、N[Q]Computing Golomb coding parameters
Figure FDA0002203011330000032
(10) Mapping the modulus value E _ mod of the residual error to be a non-negative value to obtain a residual error mapping value ME;
(11) and updating the context parameters A [ Q ], B [ Q ], C [ Q ] and N [ Q ] according to the residual modulus value E _ Q of the current pixel.
6. The method of claim 5, wherein step (2) is specifically as follows:
if D isi≤-T3Then, thenQi＝-4；
If D isi≤-T2Then Q isi＝-3；
If D isi≤-T1Then Q isi＝-2；
If D isiIf < -Near, then Qi＝-1；
If D isiLess than or equal to Near, then Qi＝0；
If D isi＜T1Then Q isi＝1；
If D isi＜T2Then Q isi＝2；
If D isi＜T3Then Q isi＝3；
else，Qi＝4。
7. The method according to claim 5, characterized in that step (10) is specifically as follows:
if ((Near ═ 0) and (k ≦ -nq)) are satisfied and (E _ mod ≧ 0), ME ≦ 2 × E _ mod + 1;
when ((Near ═ 0) and (k ═ 0) are satisfied, and (2 × B [ Q ≦ -nq ])) and (E _ mod ≧ 0) are not satisfied, ME × (E _ mod + 1);
if ((Near ═ 0) and (k ≦ 0) and (2 × B [ Q ≦ -nq ])) are not satisfied and (E _ mod ≧ 0), ME is 2 × E _ mod;
if ((Near ═ 0) and (k ≦ 0) and (2 × B [ Q ≦ -nq ])) are not satisfied and (E _ mod ≧ 0), ME is 2 × E _ mod-1.
8. The method according to claim 5, wherein step (11) is specifically as follows:
B[Q]＝B[Q]+E_q×(2×Near+1)；
A[Q]＝A[Q]+abs(E_q)；
if N [ Q ] ═ RESET is satisfied, then A [ Q ] ═ A [ Q ] > 1;
if N [ Q ] ═ RESET is satisfied and B [ Q ] ≧ 0 is satisfied, then B [ Q ] ═ B [ Q ] > 1;
if N [ Q ] ═ RESET is satisfied and B [ Q ] ≧ 0 is not satisfied, then B [ Q ] ═ - ((1-B [ Q ]) > 1), N [ Q ] > 1;
N[Q]＝N[Q]+1；
if B [ Q ] is less than or equal to-N [ Q ], then B [ Q ] + N [ Q ];
if B [ Q ] ≦ -NQ and C [ Q ] > -128 are satisfied, then C [ Q ] ═ CQ ] -1;
if B [ Q ] is not more than-NQ ] and B [ Q ] is not more than-NQ ], then B [ Q ] is-NQ + 1;
if B [ Q ] > 0 is satisfied, then B [ Q ] ═ B [ Q ] -N [ Q ];
if B [ Q ] is equal to or less than-N [ Q ] and C [ Q ] is less than 127, then C [ Q ] + 1;
if B [ Q ] is less than or equal to-N [ Q ] and B [ Q ] is more than 0, then B [ Q ] is equal to 0;
where RESET is a JPEG-LS encoding parameter, and > denotes a right shift operator.
9. The method of claim 4, wherein step S42 includes the sub-steps of:
s421, calculating Golomb coding parameters val and n according to the residual mapping value ME and the Golomb coding parameter k, wherein the calculation formula is as follows:
Figure FDA0002203011330000051
n＝ME-val×2k
s422, carrying out Golomb length-limited coding on the E according to val and n to obtain a code stream;
and S423, outputting the minimum pixel bit number bpp and the first pixel FirstPixel of the current independent coding region together as side information and the code stream of the current independent coding region.
10. A computer-readable storage medium, wherein the computer-readable storage medium stores thereon a computer program, and when executed by a processor, the computer program implements the large format remote sensing image lossless compression method according to any one of claims 1 to 9.
CN201910871670.3A 2019-09-16 2019-09-16 Large-format remote sensing image lossless compression method Active CN110691242B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201910871670.3A CN110691242B (en) 2019-09-16 2019-09-16 Large-format remote sensing image lossless compression method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201910871670.3A CN110691242B (en) 2019-09-16 2019-09-16 Large-format remote sensing image lossless compression method

Publications (2)

Publication Number Publication Date
CN110691242A true CN110691242A (en) 2020-01-14
CN110691242B CN110691242B (en) 2020-08-28

Family

ID=69109394

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201910871670.3A Active CN110691242B (en) 2019-09-16 2019-09-16 Large-format remote sensing image lossless compression method

Country Status (1)

Country Link
CN (1) CN110691242B (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2023179464A1 (en) * 2022-03-23 2023-09-28 华为技术有限公司 Image compression method, image decompression method and image processing apparatus
CN117354534A (en) * 2023-12-04 2024-01-05 上海方诚光电科技有限公司 Adaptive image lossless compression method, device and storage medium

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101478693A (en) * 2008-12-31 2009-07-08 中国资源卫星应用中心 Method for evaluating star-loaded optical remote sensing image compression quality
CN103618898A (en) * 2013-12-09 2014-03-05 中国计量学院 Complexity image lossless compression method supporting random access
CN108092669A (en) * 2017-12-28 2018-05-29 厦门大学 A kind of self-adapting data compression method and system based on discrete cosine transform
US20180182127A1 (en) * 2016-12-27 2018-06-28 Intel Corporation Near lossless compression scheme and system for processing high dynamic range (hdr) images

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101478693A (en) * 2008-12-31 2009-07-08 中国资源卫星应用中心 Method for evaluating star-loaded optical remote sensing image compression quality
CN103618898A (en) * 2013-12-09 2014-03-05 中国计量学院 Complexity image lossless compression method supporting random access
US20180182127A1 (en) * 2016-12-27 2018-06-28 Intel Corporation Near lossless compression scheme and system for processing high dynamic range (hdr) images
CN108092669A (en) * 2017-12-28 2018-05-29 厦门大学 A kind of self-adapting data compression method and system based on discrete cosine transform

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
张玉山等: "一种红外点目标图像高保真压缩方法", 《上海航天》 *
张玉山等: "卫星遥感图像无损压缩技术改进方法", 《航天器工程》 *

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2023179464A1 (en) * 2022-03-23 2023-09-28 华为技术有限公司 Image compression method, image decompression method and image processing apparatus
CN117354534A (en) * 2023-12-04 2024-01-05 上海方诚光电科技有限公司 Adaptive image lossless compression method, device and storage medium
CN117354534B (en) * 2023-12-04 2024-02-02 上海方诚光电科技有限公司 Adaptive image lossless compression method, device and storage medium

Also Published As

Publication number Publication date
CN110691242B (en) 2020-08-28

Similar Documents

Publication Publication Date Title
CN111133476B (en) System, apparatus and method for compression and decompression of a point cloud comprising a plurality of points
CN107211128B (en) Adaptive chroma downsampling and color space conversion techniques
CN104702962B (en) Decoding method, encoder and decoder in frame
CN110710217B (en) Method and apparatus for coding last significant coefficient flag
EP1977605B1 (en) Image processing
US7609899B2 (en) Image processing apparatus, image processing method, and recording medium thereof to smooth tile boundaries
JP2010514246A (en) Image compression and decompression
JP6568078B2 (en) Method and apparatus for encoding image data and method and apparatus for decoding image data
KR100922510B1 (en) Image coding and decoding method, corresponding devices and applications
US10785493B2 (en) Method of compressing and decompressing image data
CN115150612A (en) Image encoding and decoding method, encoder, decoder, and storage medium
CN115606179A (en) CNN filter for learning-based downsampling for image and video coding using learned downsampling features
US11394987B2 (en) Chroma samples from luma samples prediction for video coding
CN110691242B (en) Large-format remote sensing image lossless compression method
CN107105208A (en) A kind of lossless coding and coding/decoding method of Bayer images
US9143160B2 (en) Co-compression and co-decompression of data values
JP2017192078A (en) Picture encoder and control method thereof
CN114640857A (en) Image compression method and 3D noise reduction method based on improved Huffman coding
CN114788280A (en) Video coding and decoding method and device
US11463716B2 (en) Buffers for video coding in palette mode
US20030156651A1 (en) Method for reducing code artifacts in block coded video signals
CN110710219B (en) Method and apparatus for context derivation for coefficient coding
Naaz et al. Implementation of hybrid algorithm for image compression and decompression
CN117097905B (en) Lossless image block compression method, lossless image block compression equipment and storage medium
Visavakitcharoen et al. Compression performance of extended JPEG XT under various TMOs

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