CN108848385B

CN108848385B - Block data compression method suitable for micro-nano satellite

Info

Publication number: CN108848385B
Application number: CN201810634416.7A
Authority: CN
Inventors: 周诠; 回征; 张晨光; 呼延烺; 张茗茗; 刘娟妮; 魏佳圆
Original assignee: Xian Institute of Space Radio Technology
Current assignee: Xian Institute of Space Radio Technology
Priority date: 2018-06-20
Filing date: 2018-06-20
Publication date: 2021-04-13
Anticipated expiration: 2038-06-20
Also published as: CN108848385A

Abstract

The invention provides a block data compression method suitable for a micro-nano satellite, which changes three groups of different types of data into 4 groups of same types of data, fully utilizes the prior knowledge of the block data, reduces the transmitted data volume on the premise of not reducing the PSNR performance, and greatly improves the compression ratio. In the case of 4 times compression, the actual compression ratio is increased by 15% -33%. The invention provides a simple and efficient compression method, which is convenient for hardware implementation and can provide support for application of a micro-nano satellite.

Description

Block data compression method suitable for micro-nano satellite

Technical Field

The invention relates to a data transmission method, in particular to a method for combining image compression and hiding, belonging to the field of communication (such as data communication technology and the like).

Background

Data transmission and image processing are currently important research topics in the field of communications. With the development of science and technology, people have greater and greater requirements on high-resolution images, and data compression is imperative.

Data compression is divided into a lossy compression method and a lossless compression method, the lossy compression method is high in compression ratio, but certain information loss exists between a compressed recovery image and an original image, and the application is not greatly influenced in practice as long as subjective and objective evaluation indexes meet requirements. The lossless compression method has no information loss between the recovered image and the original image after compression, but the compression ratio is particularly small, generally about 2 times, the use occasion is limited, and data transmission is inconvenient. The compression method adopted by the general high-speed data transmission system is mostly a lossy method, and for images, the effect that the peak signal-to-noise ratio (PSNR) is more than 30dB is ideal.

There are many image data compression methods, representative of which are JPEG and JPEG 2000. The compression method is complex and difficult to realize by hardware, and the JPEG2000 compression algorithm is realized mostly based on foreign compression chips at present, has high development cost and is limited by people; or some satellite tasks do not need high image quality, such as satellite extravehicular monitoring, earth observation general survey, small satellites, micro-nano satellite image transmission, technical test satellite image transmission, space station internal monitoring, ground monitoring systems and the like, and the adoption of an overseas JPEG compression chip is not needed.

The Block Truncation Coding (BTC) compression method is a classic low-cost compression method, a typical compression ratio is 4 times, a peak signal-to-noise ratio (PSNR) is more than 30dB, and the method can be applied to micro-nano satellite occasions, and is more suitable for the occasions such as micro-nano satellites and the like requiring simple software and hardware realization and meeting the requirements on performance if the compression ratio can be improved or the compression performance is improved when the compression ratio is unchanged.

Disclosure of Invention

The technical problem solved by the invention is as follows: the method overcomes the defects of the prior art, provides a block data compression method suitable for a micro-nano satellite, changes three groups of different types of data compressed in blocks into 4 groups of data of the same type, performs compression coding based on prior knowledge of the data, reduces the data transmission amount, achieves the aim of improving the compression ratio of the block compression method without increasing the implementation complexity, and meets the requirements of users on a low-cost block compression technology.

The technical scheme of the invention is as follows: a block data compression method suitable for a micro/nano satellite comprises the following steps:

(1) assuming that the size of the original image a1 is M × N, 8 bits are quantized, and the original image a1 is divided into non-overlapping blocks of size K1 × K2, each block having X1, X2, … Xk gray values, K being K1 × K2; both K1 and K2 are positive integers;

(2) calculating the gray average value m of each non-overlapped block to obtain a set B consisting of k 1-bit values Bj, wherein j is 1,2 and … k; wherein B satisfies the condition: if Xj is larger than or equal to m, Bj is equal to 1, otherwise Bj is equal to 0; calculating the gray average value L of Xj corresponding to Bj being 0, calculating the gray average value H of Xj corresponding to Bj being 1, and obtaining three groups of numbers L, H and B for each non-overlapped block;

(3) changing three groups of numbers L, H and B of each non-overlapping block into four groups of numbers L, H, C1 and C2, wherein C1 and C2 are decimal numbers formed by any 8 bits of 1-16 bits, and performing the operation of the step (2) on each non-overlapping block of the original image A1 to finish compression;

(4) rearrange L, H, C1, C2:

when C1 is not more than C2, the arrangement is L, H, C1 and C2; as X, Y, Z1, Z2;

when C1> C2, the arrangement is H, L, C1, C2; as X, Y, Z1, Z2;

(5) and carrying out difference value coding processing on the X, Y, Z1 and Z2 to obtain processed results U, V, W1 and W2:

(6) the receiving end decompresses the data to obtain four groups of numbers L, H, C1 and C2;

(7) and arranging the data according to the sequence of the LH C1C 2 to obtain decoded data, arranging the data according to the sequence of the LHB to obtain the decoded data LHB, and recovering the original image A1.

The specific process of performing difference coding processing on X, Y, Z1, Z2 in the step (5) is as follows:

when X is less than or equal to Y, the difference value after coding treatment is as follows: H-L, H, C2-C1, C2; is marked as S, U, V, W1, W2, S is 0;

when X > Y, the difference value coding process is as follows: h, H-L, C1, C1-C2; recording S, U, V, W1, W2 and S is 1;

h, L, C1, C2 are represented by 8 bits, the number of difference bits is less than 8, and H-L is typically 3 or 4 bits.

The specific process of performing data decompression on the receiving end in the step (6) is as follows: setting the data received by the receiving end as S, U, V, W1 and W2;

if S is 0, U ≦ V, then L ═ V-U, H ═ V, W2-W1 ═ C1, W2 ═ C2;

if S is 1, U > V, then U-V ═ L, H ═ U, W1 ═ C1, W1-W2 ═ C2.

The method for forming C1 and C2 in the step 3) is as follows:

the first 1-8 bits of 1-16 bits directly form a decimal number C1, and 8 bits of 9-16 bits form a decimal number C2.

The method for forming C1 and C2 in the step 3) is as follows:

of the bits 1-16, 8 bits 1, 9,2,10,3,11,4,12 form the decimal number C1; the 8 bits 5,13,6,14,7,15,8,16 form the decimal number C2.

The method for forming C1 and C2 in the step 4) is as follows:

of the bits 1-16, 8 bits 1,2, 3,4,9,10,11,12 form the decimal number C1; the 8 bits 5,6,7,8,13,14,15,16 form the decimal number C2.

The method for forming C1 and C2 in the step 4) is as follows:

the decimal number C1 is formed by 1-8 bits before 1-16 bits, and the decimal number C2 is formed by 8 bits obtained by inverting 8 bits of 9-16 bits.

Compared with the prior art, the invention has the beneficial effects that:

(1) none of the prior art documents or patents have compressed the three sets of values (L, H, B) into four sets of values. After AMBTC compression, three sets of values (L, H, B) are obtained, L and H being integer values between 0 and 255. The prior art, although different, is based on three sets of values (L, H, B) processing.

(2) The invention also utilizes prior information (L < ═ H) of two mean values to carry out ingenious coding, reduces the bit number of transmission, improves the compression ratio, and ensures that the peak signal-to-noise ratio (PSNR) is basically unchanged.

(3) The invention fully utilizes the prior knowledge of the AMBTC compressed image, reduces the transmitted data volume on the premise of not reducing the PSNR performance basically, and improves the quality of the recovered image by the receiving end according to the prior knowledge.

(4) The method can select the bit number of the difference value C2-C1 or C1-C2 according to needs, improve the image compression ratio, obtain various BTC compressed images with different qualities, and improve the quality of recovered images under the condition that the compression ratio is not changed.

(5) The invention changes three groups of numbers into 4 groups of numbers, fully utilizes the prior knowledge of block data, reduces the transmitted data volume on the premise of not reducing the PSNR performance, and greatly improves the compression ratio. In the case of 4 times compression, the actual compression ratio is increased by 15% -33%.

The original BTC method compresses 4 × 4 image blocks, and each image block has 2 average values (8 bits each) and 16-bit binary numbers, which are 32 bits in total, and the last 16-bit data is not the same type as the first 16 bits and cannot be processed jointly. The invention arranges the rear 16 bits into two 8-bit values which are regarded as 2 decimal values, so that the rear 16 bits and the front 16 bits form one type of data, and the joint processing is convenient.

Lowest compression ratio estimation: take 4 x 4 per block as an example

L and H are changed into L and H-L, wherein L is 8 bits, and H-L is 3 or 4 bits;

c1 and C2 were changed to C1 and C2-C1, C1 was represented by 8 bits, and C2-C1 was represented by p bits (p ═ 1-8).

The total bit number of each block of the original method is 32 bits, the compression ratio is 4 times,

the new method has a total number of bits per block of 8+4+8+ p-20 + p and a compression ratio R of 4 x 32/(20+ p).

Even if the worst case p is 8, the compression ratio R is 4 × 32/(28) 4.6 times, and the compression ratio is increased by 15%.

When p is 4, the compression ratio R is 4 × 32/(24) is 5.3 times, and the compression ratio is increased by 33%.

Drawings

FIG. 1 is a flow chart of the method of the present invention.

Detailed Description

Example 1

To verify the performance of the algorithm presented herein, the experiment was simulated using 4 8-bit grayscale images of size 512 × 512. The degree of change of the image is expressed in terms of peak signal-to-noise ratio (PSNR). The compression ratio R is 4 and R is 4. K1 × K2 ═ 4 × 4.

The implementation steps of the block data compression method suitable for the micro/nano satellite are as follows:

(1) assuming that the size of the original image a1 is M × N, 8 bits are quantized, and the original image a1 is divided into non-overlapping blocks of size K1 × K2, each block having X1, X2, … Xk gray values, K being K1 × K2; both K1 and K2 are positive integers; m512, N512, K1 4, K2 4, K16

(2) Obtaining a gray average value m of each non-overlapped block to obtain a set B consisting of 16 1-bit values Bj, wherein j is 1,2 and … k; wherein B satisfies the condition: if Xj is larger than or equal to m, Bj is equal to 1, otherwise Bj is equal to 0; calculating the gray average value L of Xj corresponding to Bj being 0, and calculating the gray average value H of Xj corresponding to Bj being 1, namely obtaining three groups of numbers L, H and B for each non-overlapped block;

step 3) the method of forming C1 and C2 is as follows (one of four):

the method comprises the steps of 1, directly forming a decimal number C1 by 1-8 bits before 1-16 bits, and forming a decimal number C2 by 8 bits between 9 and 16 bits;

the method 2 comprises the following steps: of the bits 1-16, 8 bits 1, 9,2,10,3,11,4,12 form the decimal number C1; the 8 bits of 5,13,6,14,7,15,8,16 form the decimal number C2;

the method 3 comprises the following steps: of the bits 1-16, 8 bits 1,2, 3,4,9,10,11,12 form the decimal number C1; the 8 bits of 5,6,7,8,13,14,15,16 form the decimal number C2;

the method 4 comprises the steps that 1-8 bits before 1-16 bits form a decimal number C1, 8 bits obtained after 8 bits of 9-16 bits are inverted (0 is changed into 1,1 is changed into 0) form a decimal number C2;

(4) rearrange L, H, C1, C2:

when C1> C2, the arrangement is H, L, C1, C2; as X, Y, Z1, Z2;

(5) and carrying out difference value encoding processing on the X, Y, Z1 and Z2 to obtain processed results S, U, V, W1 and W2:

let the received data be X, Y, Z1, Z2

When X < ═ Y, the data format is: H-L, H, C2-C1, C2; is recorded as S, U, V, W1, W2, S is 0;

when X > Y, the data format is: h, H-L, C1, C1-C2; note S, U, V, W1, W2, S ═ 1;

e.g., 180,190,95,105, to 0,10, 190, 10,105,

e.g. 190,180,105,95, to 1, 190, 10,105, 10

Note: the number of bits used for encoding H-L, C2-C1 is reduced, and if 8 bits are used for H, L, C1 and C2, the number of bits used for H-L is reduced, and can be set to 3 or 4 bits.

When K1 × K2 ═ 4 × 4 ═ 16, R ═ 16 × 8/(4 × 4/2+16) ═ 16 × 8/(24) ═ 5.3

R is increased from 4 to 5.3, and the compression ratio is increased by 33%.

let the received data be S, U, V, W1, W2

If S is 0, then U is V, V-U is L, H is V, W2-W1 is C1, W2 is C2

If S is 1, U is V, U-V is L, H is U, W1 is C1, W1-W2 is C2

If S, U, V, W1, W2: 0,10,190, 10,105,

then, V-U-190-10-L-180, H-190, W2-W1-C1-105-10-95, W2-C2-105, and 180,190,95,105 are recovered;

if S, U, V, W1, W2: 1,190, 10,105,10,

then 190,180,105,95 are recovered from U-V-10-L-180, H-U-190, W1-C1-105, W1-W2-105-10-95;

Bmp as an example, the image is compressed by 4 times by using the conventional compression method of AMBTC, and the PSNR is calculated.

Image a1 and recovery a 1: compression recovery PSNR 33 dB;

by taking len. bmp as an example, the image is compressed by 4.6 times and 5.3 times by using the compression method of the invention, and the PSNR is calculated.

Image a1 and recovery a 1: the compression recovery PSNR is approximately equal to 33-32 dB.

The image data compression technology is widely applied to spacecrafts such as remote sensing satellites and space detectors and various satellite data transmission systems, and is certainly more widely applied in the future. The mainstream method for compressing the data on the satellite is based on the JPEG2000 algorithm, the compression ratio is mainly 4 times, but the JPEG2000 has high cost and is limited by people. The block data compression method is, for example, a BTC method, PSNR is not similar to a JPEG2000 compression method, but in many cases, PSNR meets application requirements, is simple and practical, but is low in compression ratio, and the compression ratio can only reach 4 times under the condition of compromise between performance and compression ratio, so that improvement of the compression ratio has important significance while keeping simplicity and practicability of block compression.

The method has the characteristics of low implementation complexity, strong practicability and the like, can be implemented at high speed by utilizing small-scale FPGA resources, and the actual compression ratio is improved by 15 percent at least and can reach 33 percent compared with a typical block compression method. The realization complexity is far lower than that of the standard JPEG compression method, so that the method has practical value in spacecraft engineering and low-cost image transmission systems.

The invention is not described in detail and is within the knowledge of a person skilled in the art.

Claims

1. A block data compression method suitable for a micro/nano satellite is characterized by comprising the following steps:

(4) rearrange L, H, C1, C2:

when C1> C2, the arrangement is H, L, C1, C2; as X, Y, Z1, Z2;

(5) carrying out differential coding processing on X, Y, Z1 and Z2 to obtain processed results U, V, W1 and W2; the difference value coding processing steps are as follows:

h, L, C1 and C2 are represented by 8 bits, the number of difference bits is less than 8, and H-L is 3 or 4 bits;

if S is 0, U ≦ V, L ═ V-U, H ═ V, W2-W1 ═ C1, W2 ═ C2;

if S is 1, U > V, U-V ═ L, H ═ U, W1 ═ C1, W1-W2 ═ C2;

(7) arranging the data according to the sequence of the data LH C1C 2 to obtain decoded data, arranging the data according to the sequence of the data L H B to obtain decoded data L H B, and further recovering an original image A1;

the step 3) forming the C1 and the C2 is one of the following methods:

a) 1-8 bits before 1-16 bits directly form a decimal number C1, and 8 bits between 9 and 16 bits form a decimal number C2;

b) of the bits 1-16, 8 bits 1, 9,2,10,3,11,4,12 form the decimal number C1; the 8 bits of 5,13,6,14,7,15,8,16 form the decimal number C2;

c) of the bits 1-16, 8 bits 1,2, 3,4,9,10,11,12 form the decimal number C1; the 8 bits of 5,6,7,8,13,14,15,16 form the decimal number C2;

d) the decimal number C1 is formed by 1-8 bits before 1-16 bits, and the decimal number C2 is formed by 8 bits obtained by inverting 8 bits of 9-16 bits.