WO2020258647A1 - Image reconstruction method and device - Google Patents

Abstract

An image reconstruction method and device. The method comprises: obtaining a DNA base sequence corresponding to a compressed image stored by utilizing DNA, the DNA base sequence comprising a plurality of DNA bases and/or base combinations (S11); decompressing the DNA base sequence into first image data according to a corresponding relationship between DNA bases and/or base combinations and a decompression value (S12); and inputting the first image data into an image reconstruction model, and outputting second image data by means of the image reconstruction model, the resolution of the second image being greater than that of the first image (S13). According to the method, the second image with higher pixel density, finer image quality and more details is reconstructed from the first image data, so that the requirement of higher image quality is met, and a powerful technical guarantee is provided for wide application of DNA image storage.

Description

Image reconstruction method and device Technical field

The present disclosure relates to the field of information processing technology, and in particular, to an image reconstruction method and device.

Background technique

With the development of life science and technology, as well as the cross-development of life science and other science and technology, it is possible to use genetic material deoxyribonucleic acid (DNA) as a storage medium. Digital information DNA storage refers to the storage of digital information in DNA base sequences. This technology uses a DNA synthesizer to artificially synthesize DNA for storage, and a DNA sequencer to read the stored information. With the development of the Internet, people's demand for high-definition images is getting higher and higher, especially in some special application fields, such as video surveillance, medical imaging, etc., the requirements for image accuracy are getting higher and higher. When using DNA to store images, in order to reduce the workload of synthesis or sequencing, it is often necessary to perform some compression processing on the image data. When performing image reconstruction, it is necessary to decode the DNA base sequence corresponding to the compressed image data, which is difficult to decode . In addition, the resolution of the decoded image is low, which is difficult to meet practical applications.

Summary of the invention

In order to overcome the problems in the related art, the present disclosure provides an image reconstruction method and device.

According to a first aspect of the embodiments of the present disclosure, there is provided an image reconstruction method, including:

Obtaining a DNA base sequence corresponding to the compressed image stored using DNA, the DNA base sequence including multiple DNA bases and/or base combinations;

Decompress the DNA base sequence into first image data according to the correspondence between the DNA base and/or base combination and the decompression value;

The first image data is input to an image reconstruction model, and second image data is output through the image reconstruction model, and the resolution of the second image is greater than the resolution of the first image.

In a possible implementation manner, the DNA base sequence includes a plurality of sub-image fragment sequences, and the multiple sub-image fragment sequences include a sequence generated by cutting wavelet transform coefficients of the original image of the compressed image.

In a possible implementation manner, the sub-image segment sequence further includes an index sub-sequence, and the index sub-sequence is used to store index information of the sub-image segment sequence.

In a possible implementation manner, the index information includes at least one of the following: sub-image level, pixel information of sub-image, type of sub-image, and sequence of sub-image fragments in said wavelet transform coefficient. s position.

In a possible implementation manner, the decompressing the DNA base sequence into the first image data according to the correspondence between the DNA base and/or base combination and the decompression value includes:

Parse the index sub-sequence to obtain index information of the sub-image segment sequence;

Determining the decompression value sequence corresponding to the sequence of the sub-image fragments according to the correspondence between the DNA base and/or base combination and the decompression value;

Splicing the respective decompression value sequences corresponding to the multiple sub-image fragment sequences into wavelet transform coefficients of the original image according to the index information;

Perform wavelet inverse transformation on the wavelet transform coefficients to obtain the first image data.

In a possible implementation manner, the splicing the decompression value sequences corresponding to the multiple sub-image fragment sequences into wavelet transform coefficients of the original image according to the index information includes:

Acquiring, according to the index information, decompression value sequences corresponding to sub-image fragment sequences of the same level, same pixel information, and same type;

According to the position of the sub-image segment sequence in the wavelet transform coefficient in the index information, the decompression value sequences corresponding to the multiple sub-image segment sequences are spliced into the wavelet transform coefficient of the original image.

In a possible implementation manner, before the parsing the index subsequence to obtain the index information of the sub image segment sequence, the method further includes:

Extract the index subsequence from the front and back ends of the DNA base sequence respectively according to the preset length of the index subsequence;

If the base sequence of the index subsequence extracted at the front end is different from the base sequence of the index subsequence extracted at the back end, the DNA base sequence is discarded.

In a possible implementation manner, the method further includes:

If there are N consecutive preset base markers corresponding to the same number in the DNA base sequence, the DNA base sequence appearing after the preset base marker is decoded into "N said numbers" , (N≥2).

In a possible implementation, the image reconstruction model is set to be obtained by training using multiple first sample image data and multiple second sample image data, and the first image and the second image are Different resolutions of the same image mean that the resolution of the second image is greater than the resolution of the first image.

In a possible implementation, the image reconstruction model is set to be obtained by training using multiple first sample image data and multiple second sample image data, including:

Acquiring the first sample image data and the second sample image data;

Constructing an image reconstruction model, in which training parameters are set;

Respectively input the first sample image data into the image reconstruction model to generate prediction results;

Based on the difference between the prediction result and the second sample image data, the training parameters are iteratively adjusted until the difference meets a preset requirement.

In a possible implementation, the image reconstruction model includes at least one of an adversarial neural network model, a convolutional neural network model, and a rapid and accurate super-resolution technology (RAISR) image reconstruction model.

According to a second aspect of the embodiments of the present disclosure, there is provided an image reconstruction device, including:

processor;

A memory for storing processor executable instructions;

Wherein, the processor is configured to execute the method described in any embodiment of the present disclosure.

According to a third aspect of the embodiments of the present disclosure, there is provided a non-transitory computer-readable storage medium. When instructions in the storage medium are executed by a processor, the processor can execute any of the method.

The technical solutions provided by the embodiments of the present disclosure may include the following beneficial effects: the present disclosure obtains the DNA base sequence corresponding to the compressed image stored using DNA, and according to the correspondence relationship between the DNA base and/or base combination and the decompression value, The DNA base sequence is decompressed into the first image data, and according to the pre-established image reconstruction model, the first image data is reconstructed with higher pixel density, finer image quality and more detailed first image data. Two images, so as to meet the demand for higher picture quality, and provide a strong technical guarantee for the wide application of DNA image storage.

It should be understood that the above general description and the following detailed description are only exemplary and explanatory, and cannot limit the present disclosure.

Description of the drawings

The drawings herein are incorporated into the specification and constitute a part of the specification, show embodiments in accordance with the disclosure, and together with the specification are used to explain the principle of the disclosure.

Fig. 1 is a flowchart showing an image reconstruction method according to an exemplary embodiment.

Fig. 2 is a schematic diagram showing a first-level decomposition of an image according to an exemplary embodiment.

Fig. 3 is a schematic diagram showing a three-level decomposition of an image according to an exemplary embodiment.

Fig. 4 is a flowchart showing an image reconstruction method according to an exemplary embodiment.

Fig. 5 is a block diagram showing an image reconstruction device according to an exemplary embodiment.

Fig. 6 is a block diagram showing an image reconstruction device according to an exemplary embodiment.

Detailed ways

Here, exemplary embodiments will be described in detail, and examples thereof are shown in the accompanying drawings. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings indicate the same or similar elements. The implementation manners described in the following exemplary embodiments do not represent all implementation manners consistent with the present disclosure. On the contrary, they are merely examples of devices and methods consistent with some aspects of the present disclosure as detailed in the appended claims.

In order to facilitate those skilled in the art to understand the technical solutions provided by the embodiments of the present disclosure, the following first describes the technical environment implemented by the technical solutions.

As a storage medium, DNA has many advantages compared with existing tape or hard disk storage media: First, the body of DNA is very small, and a base pair is only dozens of atoms in size. With DNA as the storage medium, the overall data is The volume will be much smaller than traditional optical discs or hard drives; the second is the high density of DNA. 1 gram of DNA is less than the size of a drop of dew on your fingertips, but it can store 700TB of data, which is equivalent to 14,000 Blu-ray discs with a capacity of 50GB, or 233 A 3TB hard drive, the latter weighs about 151 kilograms; third, the DNA is extremely stable and can be stored for a long time. As a storage medium, DNA also has its own unique rule attributes:

(1) Due to the limitation of synthesis technology, the length of the DNA sequence cannot be too long. If it is too long, it will cause difficulties in synthesis and the result of synthesis is prone to errors. Therefore, DNA base sequences need to be segmented when storing DNA;

(2) Images, especially high-definition images, have a large amount of data. If they are stored directly, the synthesis workload is increased;

(3) If there are consecutive identical base sequences in the DNA sequence, it will also cause difficulties in synthesis, and the synthesis results are prone to errors. Based on the above technical limitations, when using DNA as a storage medium for image decoding and reconstruction, the decoding is difficult and the image resolution is low.

Based on actual technical requirements similar to those described above, the present disclosure proposes an image reconstruction method and device.

The data processing method described in the present disclosure will be described in detail below with reference to FIG. 1. Fig. 1 is a method flowchart of an embodiment of an image reconstruction method provided by the present disclosure. Although the present disclosure provides method operation steps as shown in the following embodiments or drawings, more or less operation steps may be included in the method based on conventional or without creative labor. In steps where there is no necessary causality logically, the execution order of these steps is not limited to the execution order provided by the embodiments of the present disclosure.

Specifically, an embodiment of an image reconstruction method provided by the present disclosure is shown in FIG. 1. The method can be applied to image decoding and reconstruction using DNA as a storage medium, including:

Step S11, obtaining a DNA base sequence corresponding to the compressed image stored using DNA, the DNA base sequence including a plurality of DNA bases and/or base combinations.

In the embodiments of the present disclosure, the DNA base/or base combination includes: a single base constituting DNA, the bases adenine (A), guanine (G), cytosine (C) and thymine (T) ); or any base combination, such as AG, GCT, wherein the number of bases in the base combination can be two or more. The compressed image includes the use of image compression algorithms such as discrete cosine transform (DCT, Discrete Cosine Transform), wavelet transform, etc. to process the original image with storage. It should be noted that the setting of the compressed image is not limited to the above examples. Those skilled in the art may also make other changes under the enlightenment of the technical essence of this application, but as long as the functions and effects achieved are the same or similar to those of this application, they shall be covered by the protection scope of this application. Perform DNA encoding and storage on the compressed image data to obtain the DNA base sequence corresponding to the compressed image. The present disclosure can use a DNA sequencer to detect the arrangement of DNA bases and/or base combinations in the synthesized DNA base sequence order.

Step S12: Decompress the DNA base sequence into first image data according to the corresponding relationship between the DNA base and/or base combination and the decompression value.

In the embodiments of the present disclosure, the decompression value includes data obtained by compressing the original image by any one of the image compression methods in the foregoing embodiments, for example, wavelet transform coefficients obtained by image compression of the original image by wavelet transform, or DCT transforms the DCT coefficient obtained by image compression of the original image, etc. The corresponding relationship between the DNA base and/or base combination and the decompression value can be preset, and the corresponding relationship between the DNA base and/or base combination and the decompression value can be a one-to-one correspondence as shown in Table 1. Among them, in order to facilitate the calculation, the number column of Table 1 is the number after the decimal point. When encoding, the effective value of wavelet transform is reserved to two digits. It should be noted that the setting of the corresponding relationship between the DNA base and/or base combination and the decompression value is not limited to the above examples, and those skilled in the art may also make other changes under the enlightenment of the technical essence of this application. However, as long as the functions and effects achieved are the same or similar to those of this application, they should be covered by the scope of protection of this application.

In the embodiments of the present disclosure, the acquired DNA base sequence is decoded into the compressed data of the original image according to the corresponding relationship between the DNA base and/or base combination and the decompression value, and the compressed data is decompressed in reverse. After processing, the first image data is obtained. Because part of the data is discarded when the original image is encoded, the resolution of the decompressed first image data is lower than the original image. The opposite decompression processing includes: for example, using wavelet transform to compress the original image, encoding the wavelet transform coefficients into DNA base sequences, and performing inverse wavelet transform on the decoded wavelet transform coefficients during decompression to obtain the first image data.

Step S13: Input the first image data to an image reconstruction model, and output second image data via the image reconstruction model, the resolution of the second image is greater than the resolution of the first image.

In the embodiment of the present disclosure, the image reconstruction model includes pre-established by the following method: using single image super resolution (Single Image Super Resolution, SISR) technology and machine learning technology to compare low-resolution images of the same image Train and learn with high-resolution images to obtain an image reconstruction model. The first image data is input to the image reconstruction model, and the second image data is output through the image reconstruction model, and the second image data has a higher resolution than the first image data.

In the embodiment of the present disclosure, by acquiring the DNA base sequence corresponding to the compressed image stored using DNA, the DNA base sequence is decompressed into the first image data according to the correspondence relationship between the DNA base and/or base combination and the decompression value , And according to the pre-established image reconstruction model, reconstruct the second image with higher pixel density, finer image quality and more details from the first image data, so as to meet the requirements of higher image quality, It provides a strong technical guarantee for the wide application of DNA image storage.

Table 1. Correspondence between numbers and DNA base sequences

digital	DNA	digital	DNA	digital	DNA	digital	DNA	digital	DNA
0	TA	20	CAC	40	AACA	60	AGCG	80	CCGC
1	AA	twenty one	CAG	41	AACC	61	AGGA	81	CCGG
2	AC	twenty two	CCA	42	AACG	62	AGGC	82	CGAA
3	AG	twenty three	GCCA	43	AAGA	63	AGGG	83	CGAC
4	CA	twenty four	CCG	44	AAGC	64	CAAA	84	CGAG
5	CC	25	CGA	45	AAGG	65	CAAC	85	CGCA
6	CG	26	CGC	46	ACAA	66	CAAG	86	CGCC
7	GA	27	CGG	47	ACAC	67	CACA	87	CGCG
8	GC	28	GAA	48	ACAG	68	CACC	88	CGGA
9	GG	29	GAC	49	ACCA	69	CACG	89	CGGC
10	AAA	30	GAG	50	ACCC	70	CAGA	90	CGGG
11	AAC	31	GCA	51	ACCG	71	CAGC	91	GAAA
12	AAG	32	GCC	52	ACGA	72	CAGG	92	GAAC
13	ACA	33	GCG	53	ACGC	73	CCAA	93	GAAG
14	ACC	34	GGA	54	ACGG	74	CCAC	94	GACA
15	ACG	35	GGC	55	AGAA	75	CCAG	95	GACC
16	AGA	36	GCCG	56	AGAC	76	GCGA	96	GACG
17	AGC	37	GCAA	57	AGAG	77	GCAC	97	GAGA
18	AGG	38	AAAC	58	AGCA	78	GCGC	98	GAGC
19	CAA	39	AAAG	59	AGCC	79	CCGA	99	GAGG

Note: The integer "1" is represented by "TCGCCA", the integer "-1" is "TGGCCA"

In a possible implementation manner, the DNA base sequence includes a plurality of sub-image fragment sequences, and the multiple sub-image fragment sequences include a sequence generated by cutting wavelet transform coefficients of the original image of the compressed image.

In the embodiments of the present disclosure, in order to effectively store the DNA sequence, the base sequence corresponding to the compressed image is segmented into multiple sub-image fragment sequences, which can be performed by the following segmentation method: performing wavelet transform on the original image to obtain multi-level sub-images, The DNA base sequence corresponding to the target wavelet transform coefficients in the same row of the same sub-image is marked with row numbers. In one example, a continuous numbering method can be used. For example, for a three-level high-frequency level sub-image, the wavelet transform coefficient matrix The size is 650×480. For the DNA sequence encoded by the 480 wavelet transform coefficients in the first line, the line number is marked 1; the 480 wavelet transform coefficients in the second line encode the DNA sequence and the line number is marked 2; and so on, a total of 650 Line, the maximum line number is 650. According to the preset length value, the DNA base sequence of the same line number is segmented to obtain the X-segment sub-image segment sequence (X≥1). The selection of the preset length value and the preset width value is not limited to The need for DNA synthesis technology and synthesis tools. In an example, the correspondence between the row number and the DNA base sequence may include the correspondence in Table 2, and the correspondence between the segment number and the DNA base sequence may include the correspondence in Table 3. In the embodiment of the present disclosure, the sub-image fragment sequences of the same line number and different segment numbers in the same sub-image are spliced in the order of segment numbers to obtain the wavelet transform coefficients of the sub-images.

In the embodiment of the present disclosure, the process of performing wavelet transformation on the original image to obtain multi-level sub-images may include: In one example, the Mallat pyramid decomposition algorithm in wavelet transformation may be used to perform wavelet transformation on the original image: For an image with m rows and n columns, the Mallat pyramid decomposition algorithm wavelet transformation process is as shown in Figure 2. First, perform one-dimensional wavelet transformation on each row of the image to obtain low-frequency coefficients L1 and high-frequency coefficients H1, and then One-dimensional wavelet transform is performed on each column of the obtained LH image (the size is still m rows and n columns), so that the image after the first-level wavelet transform can be divided into four parts: LL1, HL1, LH1, and HH1. Among them, LL1 It is a first-level low-frequency sub-image, HL1 is a first-level high-frequency horizontal sub-image, LH1 is a first-level high-frequency vertical sub-image, and HH1 is a first-level high-frequency diagonal sub-image. As shown in Fig. 3, the second, third and even higher two-dimensional wavelet transform is to perform the first-level wavelet transform on the low-frequency sub-image LL1 of the upper-level wavelet transform image. In Figure 3, 1, 2, and 3 indicate the number of decomposition levels, that is, the level of the sub-image, L indicates the low-frequency coefficient, and H indicates the high-frequency coefficient.

Table 2 Encoding of "line number" in index mark

Line number	0	1	2	3	4	5	6	7	8	9
coding	TA	TC	TG	AT	AC	AG	CG	CT	GT	GC

Table 3 Encoding of "Segment Number" in Index Mark

Segment number	0	1	2	3	4	5	6	7	8	9
coding	TA	TC	TG	AT	AC	AG	CG	CT	GT	GC

In the embodiments of the present disclosure, considering the influence of DNA synthesis technology, the original image is compressed by means of wavelet transform, and the wavelet transform coefficients are cut and stored in units of sub-images, which is conducive to the scattered storage of DNA bases during decoding. Sequences are classified to reduce the difficulty of decoding and improve decoding efficiency.

In a possible implementation manner, the sub-image segment sequence further includes an index sub-sequence, and the index sub-sequence is used to store index information of the sub-image segment sequence.

In the embodiment of the present disclosure, the index sub-sequence is used to store index information of the sub-image segment sequence, and the index information may include the sub-image generation information, such as the level of the sub-image, the type of the sub-image, etc. It can include information related to the sub-image during encoding, such as brightness information and chrominance information in the sub-image. The establishment of the index sub-sequence is conducive to accurately knowing the information related to the sub-image and facilitates the classification and splicing of the sub-images. .

In the embodiments of the present disclosure, considering how to accurately categorize scattered sub-image segment sequences during decoding, an index sub-sequence is added to the sub-image segment sequence to play a guiding role and improve decoding efficiency.

In a possible implementation manner, the index information includes at least one of the following: sub-image level, pixel information of sub-image, type of sub-image, and sequence of sub-image fragments in said wavelet transform coefficient. s position.

In the embodiments of the present disclosure, the sub-image levels include sub-images of different levels obtained by performing wavelet transformation on the original image in the foregoing embodiment. The types of the sub-images include sub-images of the same level in the foregoing embodiment. Low-frequency sub-images, high-frequency horizontal sub-images, high-frequency vertical sub-images, and high-frequency diagonal sub-images, the position of the sequence of sub-image segments in the wavelet transform coefficients, such as the sequence of sub-image segments in the foregoing embodiment The line number and segment number.

In the embodiment of the present disclosure, the pixel information of the sub-image includes brightness information, chroma information, and saturation information of the sub-image, and the process of obtaining the pixel information of the sub-image may include: according to the matrix data of the RGB color space of the original image Each point value of the matrix data ranges from 0 to 255. The RGB data can be converted by formula (1), formula (2), and formula (3), and the data of the original image can be converted Compress to obtain the YUV color data of the original image, and the formula (1), formula (2) and formula (3) include:

Y = 0.299R+0.587G+0.114B (1)

U=-01687R-0.3313G+0.5B (2)

V=0.5R-0.4187G-0.0813B (3)

Among them, Y represents the brightness, that is, the grayscale value; and U represents the chroma, and V represents the saturation, which is used to describe the color and saturation of the image, and is used to specify the color of the pixel. It should be noted that the setting method of the pixel information of the sub-image is not limited to the above examples. Those skilled in the art may also make other changes under the enlightenment of the technical essence of this application, but as long as the realized functions and effects are consistent with the original The same or similar applications shall be covered by the protection scope of this application.

The data processing method described in the present disclosure will be described in detail below with reference to FIG. 4. Fig. 4 is a method flowchart of an embodiment of an image reconstruction method provided by the present disclosure. Although the present disclosure provides method operation steps as shown in the following embodiments or drawings, more or less operation steps may be included in the method based on conventional or without creative labor. In steps where there is no necessary causality logically, the execution order of these steps is not limited to the execution order provided by the embodiments of the present disclosure.

Specifically, an embodiment of an image reconstruction method provided by the present disclosure is shown in FIG. 4, and the step S12 includes:

Step S121: parse the index subsequence to obtain index information of the sub image segment sequence.

In the embodiment of the present disclosure, the analyzing the index subsequence may include analyzing the content of the index information according to the DNA base sequence corresponding to the preset index information. For example, the corresponding relationship between the index information and the DNA base sequence may include as shown in Table 4, where Y0 represents the low-frequency sub-image of brightness, Y50 represents the five-level high-frequency level sub-image of brightness, and Y51 represents the five-level brightness. Y52 represents the five-level high-frequency diagonal sub-image of brightness, Y40 represents the four-level high-frequency horizontal sub-image of brightness, Y41 represents the four-level high-frequency vertical sub-image of brightness, Y42 represents Brightness four-level high frequency diagonal sub-image. Y30 represents a three-level high-frequency horizontal sub-image of brightness, Y31 represents a three-level high-frequency vertical sub-image of brightness, and Y32 represents a three-level high-frequency diagonal sub-image of brightness. Y20 represents a secondary high-frequency horizontal sub-image of brightness, Y21 represents a secondary high-frequency vertical sub-image of brightness, and Y22 represents a secondary high-frequency diagonal sub-image of brightness. U0 represents the low-frequency sub-image of color, U50 represents the five-level high-frequency horizontal sub-image of color, U51 represents the five-level high-frequency vertical sub-image of color, U52 represents the five-level high-frequency diagonal sub-image of color, and U40 represents four Level high frequency horizontal sub-image, U41 represents the four-level high-frequency vertical sub-image of color, and U42 represents the four-level high-frequency diagonal sub-image of color. V0 represents the low-frequency sub-image of saturation, V50 represents the five-level high-frequency horizontal sub-image of saturation, V51 represents the five-level high-frequency vertical sub-image of saturation, and V52 represents the five-level high-frequency diagonal sub-image of saturation. The four-level high-frequency horizontal sub-image of V40 saturation, V41 is the four-level high-frequency vertical sub-image of saturation, and V42 represents the four-level high-frequency diagonal sub-image of saturation. It should be noted that the setting method of the corresponding relationship between the index information and the DNA base sequence is not limited to the above examples, and those skilled in the art may also make other changes under the enlightenment of the technical essence of this application, but as long as they are implemented The functions and effects are the same or similar to those of this application, and should be covered by the scope of protection of this application.

Table 4 "Sub-image information" coding in the index mark

Y	Y0	Y51	Y51	Y52	Y40	Y41	Y42
Y code	AAT	ACG	AGA	AGT	ACA	ACT	ACC
Y	Y30	Y31	Y32	Y20	Y21	Y22	To
Y code	TGA	ATC	ATG	AAC	AAG	ATA	To
U	U0	U50	U51	U52	U40	U41	U42
U encoding	AGC	TAC	TAG	TTA	AGG	TGT	TAT
V	V0	V50	V51	V52	V40	V41	V42
V code	TGC	TCT	TCC	TCG	TTC	TTG	TCA

Step S122: Determine the decompression value sequence corresponding to the sequence of the sub-image fragments according to the corresponding relationship between the DNA base and/or base combination and the decompression value;

In the embodiments of the present disclosure, the corresponding relationship between the DNA base and/or base combination and the decompression value may be as shown in Table 1 according to the one-to-one relationship shown in Table 1, and the sub-image fragment The sequence is decoded into a sequence of decompression values. In a possible implementation, in order to reduce the total amount of correspondence settings and to distinguish between different decompression values, for example, if the DNA base sequence is "GACG", the " "GACG" is decoded to "96", or the first half of "GACG" is decoded to 7, and "CG" is decoded to 6, you can add a preset symbol mark between different decompression values, the symbol mark It preferably includes positive and negative signs. For example, when it encounters "TG", it means that "TG" is followed by a negative number; when it encounters "TC", it means that "TC" is followed by a positive number, and "TC" Or "TG" can only be decoded into the same decompression value.

Step S123, according to the index information, splice the decompression value sequences corresponding to the multiple sub-image fragment sequences into wavelet transform coefficients of the original image;

In the embodiments of the present disclosure, according to the index information, the sub-image segment sequences are classified while decoding, and the decompression values corresponding to the sub-image segment sequences of the same attribute are put together, and all the sub-image segment sequences can be unified first Decode, and then classify the decoded data according to the index information, that is, put together the decompression values corresponding to the sub-image segment sequence of the same attribute, and according to the index information, the decompression value corresponding to the sub-image segment sequence is in the wavelet transform coefficient Corresponding to the position of the wavelet transform coefficient.

Step S124: Perform inverse wavelet transform on the wavelet transform coefficients to obtain the first image data.

In the embodiments of the present disclosure, the wavelet transform coefficients that have been classified are subjected to inverse wavelet transform processing. In an example, if a certain level of sub-image data is lost or not compressed, the wavelet transform of this level of sub-image The coefficients are zero-filled, combined with the decoded data, and subjected to wavelet inverse transform processing to obtain the first image data.

In the embodiment of the present disclosure, the corresponding relationship of the corresponding DNA base sequence is established for the index information, which solves the coding problem of the index information. At the same time, based on the corresponding relationship between the index information and the DNA base sequence, the index subsequence can be compared Decoding is clever and easy to implement. When decoding, by adding a symbol coding sequence, the complexity of the correspondence between DNA bases and/or base combinations and decompression values can be reduced, and each decompression value can be distinguished, reducing the complexity of decoding.

In a possible implementation manner, the step S123, according to the index information, stitches the decompression value sequences corresponding to the multiple sub-image fragment sequences into the wavelet transform coefficients of the original image, including steps S1231 and S1232,

Step S1231: Obtain decompression value sequences corresponding to sub-image fragment sequences of the same level, same pixel information, and same type according to the index information;

Step S1232, according to the position of the sub-image segment sequence in the wavelet transform coefficients in the index information, splice the decompressed value sequences corresponding to the multiple sub-image segment sequences into wavelet transform coefficients of the original image.

In the embodiments of the present disclosure, the decompression values corresponding to the sequence of sub-image fragments of the same level can be put together according to the level information in the index information. Further, for sub-images of the same level and different types, the sub-image types are Information is classified. Furthermore, sub-images with the same level, the same type, and different pixel information are classified according to the pixel information, and finally the sub-image fragment sequences of the same level, the same type, the same pixel information, and the same row number are corresponding The decompression value of, is spliced according to the sequence of segment numbers, and the sequence of decompression values corresponding to the spliced sequence of sub-image fragments is combined into coefficients of wavelet transform according to the sequence of line numbers. In an example, if the decompression value corresponding to the sub-image segment sequence is lost, the wavelet transform coefficient at the corresponding position is set to zero.

In a possible implementation manner, in step S121, the index subsequence is parsed to obtain index information of the sub image segment sequence. It also includes step S125 and step S126 before,

Step S125: Extract the index subsequence from the front and back ends of the DNA base sequence according to the preset length of the index subsequence;

Step S126: If the base sequence of the index subsequence extracted at the front end is different from the base sequence of the index subsequence extracted at the back end, the DNA base sequence is discarded.

In the embodiment of the present disclosure, the same index subsequence is added to the front and back ends of the DNA fragment, respectively. When performing image decoding, after reading the index subsequence at the front end of the DNA fragment for the first time, when reading the index subsequence at the back end of the DNA fragment again, if the index subsequence read twice is found If they are not consistent, it means that an error occurred during the synthesis of the DNA fragment. Then the DNA fragment is discarded and not used, and the correct synthesis of the DNA fragment is searched for. Generally, the same fragment will be synthesized multiple times when the DNA fragment is synthesized. In a possible implementation manner, the base sequence corresponding to the index mark is added to the front and back ends of the DNA fragment respectively, as shown in Table 5.

Table 5 Nucleic acid fragment structure

In the embodiment of the present disclosure, index subsequences are added to the front and back ends of the DNA fragment, and the index subsequences are the same, which can play a role of verification and ensure accurate decoding.

In a possible implementation manner, if N consecutive preset base markers corresponding to the same number appear in the DNA base sequence, then the DNA base sequence that subsequently appears after the preset base marker Decode into "N said numbers", (N≥2).

In the embodiments of the present disclosure, the same number will appear N consecutive times in the synthetic DNA sequence. For example, in high-level wavelet coefficients, the probability of consecutive zeros is greater, and consecutively encoding the same DNA base sequence will cause difficulties in synthesis. , And coding redundancy, therefore, the following format "marker + value N corresponding DNA base sequence" can be used for coding. For example, the tag "TAA" represents "multiple 0s", and the wavelet coefficient "0000000000" can be represented as "TAA+AC", that is, "TAAAC", where TAA represents multiple 0s, and AC represents 10 in Table 1. When decoding, if you read the preset base tag "TAA", it will be interpreted as multiple 0s according to the preset meaning, and then read the sequence "AC" after "TAA" again. According to Table 1, it means 10, and finally decode Into 10 zeros, which is "0000000000".

The embodiments of the present disclosure take into account the difficulty of synthesizing consecutive identical base sequences. For numbers that appear consecutively, specific markers are designed to avoid the sequence synthesis of multiple consecutive identical bases and greatly reduce the synthesis of DNA base sequences The length enables fast decoding.

In a possible implementation, the image reconstruction model is set to be obtained by training using multiple first sample image data and multiple second sample image data, and the first image and the second image are Different resolutions of the same image mean that the resolution of the second image is greater than the resolution of the first image.

In the embodiment of the present disclosure, in an example, the second image sample data may adopt original image data, and perform wavelet transformation on the original image data. As in the above embodiment, a low-frequency sub-image is obtained, and the low-frequency sub-image is Perform interpolation processing until it is the same size as the corresponding second image to obtain the first image sample data. In another example, the second image sample data may use the original image data to sample the second image to obtain the intermediate Small image, and then perform interpolation processing on the middle small image to obtain the first image sample data.

In the embodiments of the present disclosure, training to obtain an image reconstruction model using multiple first sample image data and multiple second sample image data may include: establishing the image by learning the mapping relationship from the first image to the second image Reconstruction model. In one example, the image reconstruction model includes three layers, and the functions are: image block extraction, nonlinear mapping, and reconstruction. Regarding the extraction of image blocks, a series of pre-trained filters can be used to separately convolve the input low-resolution first image to obtain corresponding feature maps. The functions of the filters are different, and the functions may include, for example, input For edge detection and texture extraction in different directions of the image, any one of PCA, Haar, DCT can be used as the above-mentioned filter; for nonlinear mapping, the feature map of each image block in the first layer is nonlinearly mapped to high In the resolution tiles; regarding reconstruction, the high-resolution image blocks are grouped together to form a high-resolution image similar to the second image sample.

The embodiments of the present disclosure take into account the problem of reduced sharpness of images decoded by DNA. By adding image reconstruction models, low-resolution image data is reconstructed into high-resolution image data, and compressed images stored for DNA can be applied Provides strong technical support, improves the encoding compression rate, and reduces the synthesis workload.

In a possible implementation, the image reconstruction model is set to be obtained by training using multiple first sample image data and multiple second sample image data, including:

Step S21, acquiring the first sample image data and the second sample image data;

Step S22, construct an image reconstruction model, in which training parameters are set;

Step S23, respectively inputting the first sample image data into the image reconstruction model to generate prediction results;

Step S24, based on the difference between the prediction result and the second sample image data, iteratively adjust the training parameters until the difference meets a preset requirement.

In the embodiment of the present disclosure, in step S21, the resolution of the second image needs to be greater than the resolution of the first image. Therefore, a high-definition original image can be selected as the second sample image data {X _i }, and the The original image is processed, such as wavelet transform or DCT transform, to obtain low-resolution image data as the first sample image data {Y _i }. In step S22, an image reconstruction model F(Y; θ) is constructed, Y represents the first image, and θ represents the training parameters. In an example, the convolutional neural network parameters are used to build the image super-resolution reconstruction model, and the training Parameter θ? {W ₁ ,W ₂ ,W ₃ ,B ₁ ,B ₂ ,B ₃ }, W ₁ represents one-layer filter, B ₁ represents deviation, W ₂ represents two-layer filter, B ₂ represents mapping vector, W ₃ represents Three-layer filter, B ₃ represents a vector. Input the first sample image data {Y _i } into the image reconstruction model F(Y; θ), and the obtained prediction result Z=F(Y; θ), when the prediction result Z and its corresponding first When the loss between the two images X is the smallest, we can obtain the training parameter θ.

In the embodiments of the present disclosure, the mean square error (MSE) may be used to represent the loss function (Loss Function), referring to formula (4), n represents the number of training samples.

In an example, in order to minimize the loss, we will use stochastic gradient descent in backpropagation. The weight matrix update formulas are as equations (5) and (6), where the number of layers σ∈{1,2,3} ; I is the iterative index of this layer, δ is the step size, the initialization of the filter weight of each layer can be randomly given by Gaussian distribution with the mean value of 0, the standard of 0.001, and the deviation of 0. In an example, because in the third layer, the smaller the step size, the easier it is to converge, so in the first and second layers, δ is set to 10 ^-4 , and in the third layer, δ is set to 10 ^-5 .

In a possible implementation, the image reconstruction model includes at least one of an adversarial neural network model, a convolutional neural network model, and a rapid and accurate super-resolution technology (RAISR) image reconstruction model.

In the embodiment of the present disclosure, the confrontation neural network model includes a generation network and a discrimination network. The generation network is used to receive a random noise, and an image is generated through this noise, denoted as G(Z), and the discrimination network is used for To determine the authenticity of a picture, input an image X, output D(X), D(X) is used to represent the probability that X is a real image, and train the generation network and the discrimination network until D(G( Z)) Reach a set threshold. When the threshold is 0.5, it indicates an ideal state. The trained adversarial neural network may be used to reconstruct the low-resolution first image to generate a high-resolution second image; the convolutional neural network model may include a deep learning-based convolutional neural network and An image reconstruction model combining single-frame image super-resolution reconstruction (SISR); the precise super-resolution technology (RAISR, Rapid and Accurate Super Image Resolution) model may include the use of a series of paired low-resolution images, high Resolution image training to find filters that can be selectively applied to each pixel in low-resolution images to generate high-resolution images.

Fig. 5 is a block diagram showing a device 800 for image reconstruction according to an exemplary embodiment. For example, the device 800 may be a mobile phone, a computer, a digital broadcasting terminal, a messaging device, a game console, a tablet device, a medical device, a fitness device, a personal digital assistant, etc.

5, the device 800 may include one or more of the following components: a processing component 802, a memory 804, a power supply component 806, a multimedia component 808, an audio component 810, an input/output (I/O) interface 812, a sensor component 814, And the communication component 816.

The processing component 802 generally controls the overall operations of the device 800, such as operations associated with display, telephone calls, data communications, camera operations, and recording operations. The processing component 802 may include one or more processors 820 to execute instructions to complete all or part of the steps of the foregoing method. In addition, the processing component 802 may include one or more modules to facilitate the interaction between the processing component 802 and other components. For example, the processing component 802 may include a multimedia module to facilitate the interaction between the multimedia component 808 and the processing component 802.

The memory 804 is configured to store various types of data to support operations in the device 800. Examples of such data include instructions for any application or method operating on the device 800, contact data, phone book data, messages, pictures, videos, and so on. The memory 804 can be implemented by any type of volatile or nonvolatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable Programmable Read Only Memory (EPROM), Programmable Read Only Memory (PROM), Read Only Memory (ROM), Magnetic Memory, Flash Memory, Magnetic Disk or Optical Disk.

The power supply component 806 provides power to various components of the device 800. The power supply component 806 may include a power management system, one or more power supplies, and other components associated with generating, managing, and distributing power for the device 800.

The multimedia component 808 includes a screen that provides an output interface between the device 800 and the user. In some embodiments, the screen may include a liquid crystal display (LCD) and a touch panel (TP). If the screen includes a touch panel, the screen may be implemented as a touch screen to receive input signals from the user. The touch panel includes one or more touch sensors to sense touch, sliding, and gestures on the touch panel. The touch sensor may not only sense the boundary of a touch or slide action, but also detect the duration and pressure related to the touch or slide operation. In some embodiments, the multimedia component 808 includes a front camera and/or a rear camera. When the device 800 is in an operation mode, such as a shooting mode or a video mode, the front camera and/or the rear camera can receive external multimedia data. Each front camera and rear camera can be a fixed optical lens system or have focal length and optical zoom capabilities.

The audio component 810 is configured to output and/or input audio signals. For example, the audio component 810 includes a microphone (MIC), and when the device 800 is in an operation mode, such as a call mode, a recording mode, and a voice recognition mode, the microphone is configured to receive external audio signals. The received audio signal may be further stored in the memory 804 or transmitted via the communication component 816. In some embodiments, the audio component 810 further includes a speaker for outputting audio signals.

The I/O interface 812 provides an interface between the processing component 802 and a peripheral interface module. The peripheral interface module may be a keyboard, a click wheel, a button, and the like. These buttons may include but are not limited to: home button, volume button, start button, and lock button.

The sensor component 814 includes one or more sensors for providing the device 800 with various aspects of status assessment. For example, the sensor component 814 can detect the open/close state of the device 800 and the relative positioning of components. For example, the component is the display and the keypad of the device 800. The sensor component 814 can also detect the position change of the device 800 or a component of the device 800. , The presence or absence of contact between the user and the device 800, the orientation or acceleration/deceleration of the device 800, and the temperature change of the device 800. The sensor component 814 may include a proximity sensor configured to detect the presence of nearby objects when there is no physical contact. The sensor component 814 may also include a light sensor, such as a CMOS or CCD image sensor, for use in imaging applications. In some embodiments, the sensor component 814 may also include an acceleration sensor, a gyroscope sensor, a magnetic sensor, a pressure sensor or a temperature sensor.

The communication component 816 is configured to facilitate wired or wireless communication between the device 800 and other devices. The device 800 can access a wireless network based on a communication standard, such as WiFi, 2G, or 3G, or a combination thereof. In an exemplary embodiment, the communication component 816 receives a broadcast signal or broadcast related information from an external broadcast management system via a broadcast channel. In an exemplary embodiment, the communication component 816 further includes a near field communication (NFC) module to facilitate short-range communication. For example, the NFC module can be implemented based on radio frequency identification (RFID) technology, infrared data association (IrDA) technology, ultra-wideband (UWB) technology, Bluetooth (BT) technology and other technologies.

In an exemplary embodiment, the apparatus 800 may be implemented by one or more application specific integrated circuits (ASIC), digital signal processors (DSP), digital signal processing equipment (DSPD), programmable logic devices (PLD), field programmable A gate array (FPGA), controller, microcontroller, microprocessor, or other electronic components are implemented to implement the above methods.

In an exemplary embodiment, there is also provided a non-transitory computer-readable storage medium including instructions, such as the memory 804 including instructions, which can be executed by the processor 820 of the device 800 to complete the foregoing method. For example, the non-transitory computer-readable storage medium may be ROM, random access memory (RAM), CD-ROM, magnetic tape, floppy disk, optical data storage device, etc.

Fig. 6 is a block diagram showing a device 1900 for image reconstruction according to an exemplary embodiment. For example, the device 1900 may be provided as a server. 6, the apparatus 1900 includes a processing component 1922, which further includes one or more processors, and a memory resource represented by a memory 1932 for storing instructions executable by the processing component 1922, such as application programs. The application program stored in the memory 1932 may include one or more modules each corresponding to a set of instructions. In addition, the processing component 1922 is configured to execute instructions to perform the above-described methods.

The device 1900 may also include a power component 1926 configured to perform power management of the device 1900, a wired or wireless network interface 1950 configured to connect the device 1900 to the network, and an input output (I/O) interface 1958. The device 1900 can operate based on an operating system stored in the memory 1932, such as Windows ServerTM, Mac OS XTM, UnixTM, LinuxTM, FreeBSDTM or the like.

In an exemplary embodiment, there is also provided a non-transitory computer-readable storage medium including instructions, such as the memory 1932 including instructions, which may be executed by the processing component 1922 of the device 1900 to complete the foregoing method. For example, the non-transitory computer-readable storage medium may be ROM, random access memory (RAM), CD-ROM, magnetic tape, floppy disk, optical data storage device, etc.

Those skilled in the art will easily think of other embodiments of the present disclosure after considering the specification and practicing the invention disclosed herein. This application is intended to cover any variations, uses, or adaptive changes of the present disclosure, which follow the general principles of the present disclosure and include common knowledge or conventional technical means in the technical field not disclosed in the present disclosure . The description and the embodiments are to be regarded as exemplary only, and the true scope and spirit of the present disclosure are pointed out by the following claims.

It should be understood that the present disclosure is not limited to the precise structure that has been described above and shown in the drawings, and various modifications and changes can be made without departing from its scope. The scope of the present disclosure is limited only by the appended claims.

Claims (13)

1. An image reconstruction method, characterized in that it comprises:

   Obtaining a DNA base sequence corresponding to the compressed image stored using DNA, the DNA base sequence including multiple DNA bases and/or base combinations;

   Decompress the DNA base sequence into first image data according to the correspondence between the DNA base and/or base combination and the decompression value;

   The first image data is input to an image reconstruction model, and second image data is output through the image reconstruction model, and the resolution of the second image is greater than the resolution of the first image.
2. The method according to claim 1, wherein the DNA base sequence includes a plurality of sub-image fragment sequences, and the plurality of sub-image fragment sequences include those generated by cutting wavelet transform coefficients of the original image of the compressed image. sequence.
3. The method according to claim 2, wherein the sub-image segment sequence further includes an index sub-sequence, and the index sub-sequence is used to store index information of the sub-image segment sequence.
4. The method according to claim 3, wherein the index information includes at least one of the following: sub-image level, pixel information of the sub-image, type of the sub-image, and the sequence of the sub-image fragments in the The position in the wavelet transform coefficients.
5. The method according to claim 3, wherein the decompressing the DNA base sequence into the first image data according to the corresponding relationship between the DNA base and/or base combination and the decompression value comprises:

   Parse the index sub-sequence to obtain index information of the sub-image segment sequence;

   Determining the decompression value sequence corresponding to the sequence of the sub-image fragments according to the correspondence between the DNA base and/or base combination and the decompression value;

   Splicing the respective decompression value sequences corresponding to the multiple sub-image fragment sequences into wavelet transform coefficients of the original image according to the index information;

   Perform wavelet inverse transformation on the wavelet transform coefficients to obtain the first image data.
6. The method according to claim 5, wherein the splicing the decompression value sequences corresponding to the multiple sub-image fragment sequences into wavelet transform coefficients of the original image according to the index information comprises:

   Acquiring, according to the index information, decompression value sequences corresponding to sub-image fragment sequences of the same level, same pixel information, and same type;

   According to the position of the sub-image segment sequence in the wavelet transform coefficient in the index information, the decompression value sequences corresponding to the multiple sub-image segment sequences are spliced into the wavelet transform coefficient of the original image.
7. The method according to claim 5, characterized in that, before said analyzing said index subsequence to obtain index information of said sub image segment sequence, the method further comprises:

   Extract the index subsequence from the front and back ends of the DNA base sequence respectively according to the preset length of the index subsequence;

   If the base sequence of the index subsequence extracted at the front end is different from the base sequence of the index subsequence extracted at the back end, the DNA base sequence is discarded.
8. The method according to claim 7, further comprising:

   If there are N consecutive preset base markers corresponding to the same number in the DNA base sequence, the DNA base sequence appearing after the preset base marker is decoded into "N said numbers" , (N≥2).
9. The method according to claim 1, wherein the image reconstruction model is set to be trained using a plurality of first sample image data and a plurality of second sample image data, the first image and the The second image is a representation of different resolutions of the same image, and the resolution of the second image is greater than the resolution of the first image.
10. The method according to claim 9, wherein the image reconstruction model is set to be obtained by training using multiple first sample image data and multiple second sample image data, comprising:

    Acquiring the first sample image data and the second sample image data;

    Constructing an image reconstruction model, in which training parameters are set;

    Respectively input the first sample image data into the image reconstruction model to generate prediction results;

    Based on the difference between the prediction result and the second sample image data, the training parameters are iteratively adjusted until the difference meets a preset requirement.
11. The method according to claim 1, wherein the image reconstruction model includes at least one of an adversarial neural network model, a convolutional neural network model, and a rapid and accurate super-resolution technology (RAISR) image reconstruction model Kind.
12. An image reconstruction device, characterized in that it comprises:

    processor;

    A memory for storing processor executable instructions;

    Wherein, the processor is configured to execute the method of any one of claims 1-10.
13. A non-transitory computer-readable storage medium, when the instructions in the storage medium are executed by a processor, so that the processor can execute the method according to any one of claims 1 to 10.

Images