CN101778284A - Lossy compression method for multifocal multiphoton microscopic imaging data - Google Patents

Abstract

The invention relates to a lossy compression method for multifocal multiphoton microscopic imaging data. The existing data compression effects are not good. The method of the invention comprises the following steps: extracting binary format imaging data from an imaging data file and carrying out lossy compression on the extracted image data by a JPEG algorithm. The extracted binary format data file forms an image matrix; any layer of the image matrix is compressed by the JPEG algorithm; firstly, the image of the layer is divided into nonoverlapping 8*8 subimages; then the subimages are coded by carrying out grey level moving, forward discrete cosine transformation, normalization, quantization and Huffman coding on the subimages; and finally, all the subimages of the layer are coded by the method. The invention takes example by the characteristic of a JPEG coding standard that the information which has less effect on the image quality is neglected, carries out lossy compression on a multifocal multiphoton microscopic imaging image data block, and obtains considerable compressibility.

Description

A kind of lossy compression method method of imaging data of multifocal multiphoton microscope

Technical field

The invention belongs to the multifocal multiphoton microscopic imaging field, relate to a kind of lossy compression method method of imaging data of multifocal multiphoton microscope.

Background technology

Multifocal multiphoton microscope is one of the most effective biological tissue three-dimensional imaging technology in the world today.This technology is to adopt laser as light source on traditional fluorescence microscope imaging basis, installed laser scanning device additional, converge in the focal position with near-infrared laser and make it to realize that multiphoton fluorescence excites, utilize multiphoton excitation to reach the effect that weakens photodamaged and photobleaching simultaneously, and then the fluoroscopic image of the high-resolution cell or tissue interior detail micro-structural that obtains, and utilize the image processing techniques of computer, make the researcher can observation of cell or the fine structure and the metamorphosis of organization internal.Multifocal multiphoton microscope has become important research means in the fields such as morphology, molecular cytobiology, Neuscience, tumor research and the detection of optics live body at present.

Because the more traditional two-photon microscope of multifocal multiphoton microscope has that image taking speed is faster, the outstanding effect advantage of imaging depth and contrast.Yet this development has also caused the micro-image data volume of its imaging to become more and more huger.The liver specimens imaging data of being quoted as embodiment part in this specification only data block of place scanning just reaches 4MB, carries out the data volume that complete scan detects and will reach the GB magnitude.So huge data volume give image storage, transmit and the technology that reads has proposed stern challenge, one of key technology that solves this class problem is exactly the lossy compression of image.The lossy compression method of image can make image under the prerequisite that guarantees certain mass, and existing expression of image information is transformed into the expression-form that another can make the comparatively considerable minimizing of data volume acquisition.

Summary of the invention

The object of the present invention is to provide a kind of lossy compression method method of imaging data of multifocal multiphoton microscope.

The lossy compression method method of imaging data of multifocal multiphoton microscope of the present invention comprises the imaging data that extracts binary format from the imaging data file and adopts jpeg algorithm that the view data of extracting is carried out lossy compression method that concrete steps are:

Step (1) reads the binary format data file of multifocal multiphoton microscopic imaging, and making this binary format data file generate resolution is that M * N, each pixel are 8, the image of L layer altogether.M, N and L are by the scanning imagery characteristic decision of multifocal multiphoton microscope.

The binary format data file forms the image array of the M * N resolution of L layer, and it is stored among the three-dimensional matrice img (be the formed layering image of imaging data block, its size is M * N * L, and the element span is 0～255).

Step (2) extract the i layer data of three-dimensional matrice img as the input f of compression (x, y), 1≤i≤L wherein, adopt jpeg algorithm come to f (x y) compresses, and its concrete steps are as follows:

1. (x y) is divided into nonoverlapping 8 * 8 subimage, and j piece subimage is f with f ₈(x, y), its by

2. (x, y) each pixel of all subimages in from left to right deducts 2 from top to bottom to f ^K-1Do gray scale and move, obtain the subimage f ' after j piece subimage gray scale moves ₈(x, y), wherein k is the bit number of image;

3. the subimage f ' after j piece subimage gray scale being moved ₈(x y) carries out Forward Discrete Cosine Transform (DCT) conversion, obtains f ' ₈(x, Forward Discrete Cosine Transform T y) (u, v),

$T(u,v)=\underset{x=0}{\overset{7}{\mathrm{\Σ}}}\underset{y=0}{\overset{7}{\mathrm{\Σ}}}{f}_8^{\′}(x,y)\mathrm{\α}\left(u\right)\mathrm{\α}\left(v\right)\cos \left[\frac{(2x+1)\mathrm{u\π}}{16}\right]\cos \left[\frac{(2y+1)\mathrm{v\π}}{16}\right]$

u＝0，1，2，…7，v＝0，1，2，…7 $\mathrm{\α}\left(u\right)=\left\{\begin{array}{cc}\sqrt{\frac{1}{8}}& u=0\\ \frac{1}{2}& u=\mathrm{1,2},...,7\end{array}\right.;$ $\mathrm{\α}\left(u\right)=\left\{\begin{array}{cc}\sqrt{\frac{1}{8}}& u=0\\ \frac{1}{2}& u=\mathrm{1,2},...,7\end{array}\right.$

4. (u v) carries out normalization and quantification and obtains with T

u＝0，1，…，7 v＝0，1，…，7

In the formula, round rounds up to get the function of nearest integer, and (u v) is a luminance component quantization table in the Joint Photographic Experts Group Annex K item to Z;

5. will

Rearrange element position according to the Z-shaped pattern in the Joint Photographic Experts Group, obtain an one-dimension array that 64 elements are arranged, to first element DC coefficient in this one-dimension array

Carry out huffman coding with other elements A C coefficient that removes first element in this one-dimension array; Ask for j piece subimage in the i tomographic image

And the DC difference coefficient DC between the j-1 piece subimage DC coefficient _Diff,

${\mathrm{DC}}_{\mathrm{diff}}={\hat{T}}_j\left(\mathrm{0,0}\right)-{\hat{T}}_{j-1}\left(\mathrm{0,0}\right)$ In the formula,

JPEG coefficient coding classification chart among the Annex K is determined j piece subimage difference coefficient DC in the contrast Joint Photographic Experts Group _DiffAffiliated DC difference class, contrast the DC coefficient luminance component huffman coding table of giving tacit consent among the Joint Photographic Experts Group Annex K then and inquire about j piece subimage difference coefficient DC _DiffBasic coding and coding total length value;

When coding total length value during, but deduct basic coding length value value K, with j piece subimage difference coefficient Dc with the total length value of encoding greater than the basic coding length value _DiffLeast significant bit be arranged in the binary number of K bit and subtract 1, obtain the additional coding of binary number at last;

Basic coding and additional coded sequence are combined into j piece subimage difference coefficient DC _DiffThe coding of variable-length;

6. non zero AC coefficient in the one-dimension array of step in is 5. encoded one by one; JPEG coefficient coding classification chart among the Annex K is determined the AC class that non zero AC coefficient is affiliated one by one in the contrast Joint Photographic Experts Group;

According to the number of null value before aforementioned AC generic data that inquire and the non zero AC coefficient, contrast basic coding and the coding total length value that the AC coefficient luminance component huffman coding table of giving tacit consent among the Joint Photographic Experts Group Annex K is inquired about each non zero AC coefficient with these two data;

When coding total length value during, but deduct basic coding length value value K, the least significant bit of AC coefficient is arranged in the binary number of K bit and subtracts 1, obtain the additional coding of binary number at last with the total length value of encoding greater than the basic coding length value;

Basic coding and additional coded sequence are combined into the coding of the variable-length of non zero AC coefficient;

7. the coefficient that step is null value to the end item in the one-dimension array in is 5. represented with EOB sign EOB, the AC coefficient luminance component huffman coding table inquiry EOB that gives tacit consent among the contrast Joint Photographic Experts Group Annex K is encoded to 1010, and the coding of EOB is added into step coding end 6., finish the coding of j piece subimage in the i tomographic image;

8. according to step method 1.～7., ask for the coding of other subimage blocks of i tomographic image;

Step (3) is compressed other layers among the three-dimensional matrice img according to the method for step (2), finishes the lossy compression method of imaging data of multifocal multiphoton microscope.

The present invention's employing is encoded the multifocal multiphoton microscopic imaging view data and is formed the purpose that new data file reaches packed data by Joint Photographic Experts Group, its biggest advantage is to have used for reference ignoring of the JPEG coding standard information feature little to picture quality influence, compression multifocal multiphoton microscopic imaging video data block makes initial data descend to some extent in picture quality but influences under the little situation and obtained considerable compression ratio obtaining image information itself with diminishing.

Embodiment

The lossy compression method method of imaging data of multifocal multiphoton microscope, the step of the following concrete enforcement of employing on the MATLAB6.5 programming platform:

Step (1) is used the binary format data file gfp_106000.int that fopen () function is opened, fread () function reads multifocal multiphoton microscopic imaging, makes that this binary format data file generates that resolution is 192 * 192, each pixel is 8, totally 30 layers image.

The binary format data file forms the image array of 30 layers 192 * 192 resolution, and it is stored among the three-dimensional matrice img (be the formed layering image of imaging data block, its size is 192 * 192 * 30, and the element span is 0～255).

The 1st layer data that step (2) is extracted three-dimensional matrice img as the used input f of compression (x, y) (be f (x, y)=img (::, 1)), adopt jpeg algorithm come to f (x y) compresses, and its concrete steps are as follows:

1. (x y) is divided into nonoverlapping 8 * 8 subimage, and j piece subimage is f with f ₈(x, y), wherein

2. (x, y) each pixel of all subimages in from left to right deducts 2 from top to bottom to f ^8-1=128 do gray scale moves, and obtains the subimage f ' after j piece word image gray scale moves ₈(x, y), wherein k=8 is the bit number of image;

3. use the subimage f ' after dctmtx () and blkproc () function move j piece word image gray scale ₈(x y) carries out dct transform, obtains f ' ₈(x, and Forward Discrete Cosine Transform T y) (u, v);

4. (u v) uses the luminance component quantization table of giving tacit consent among the Joint Photographic Experts Group Annex K and blkproc () function carries out normalization and quantification obtains with T

5. will

Rearrange element position according to the Z-shaped pattern in the Joint Photographic Experts Group, obtain an one-dimension array that 64 elements are arranged, to first element DC coefficient in this one-dimension array Carry out huffman coding with other elements A C coefficient that removes first element in this one-dimension array; Ask for j piece subimage in the i tomographic image

And the DC difference coefficient DC between the subimage DC coefficient of front _Diff,

${\mathrm{DC}}_{\mathrm{diff}}={\hat{T}}_j\left(\mathrm{0,0}\right)-{\hat{T}}_{j-1}\left(\mathrm{0,0}\right)$

In the formula, 1≤j≤576,

JPEG coefficient coding classification chart among the Annex K is determined j piece subimage difference coefficient DC in the contrast Joint Photographic Experts Group _DiffAffiliated DC difference class, the DC coefficient luminance component huffman coding table of giving tacit consent among the contrast Joint Photographic Experts Group Annex K is inquired about j piece subimage difference coefficient DC _DiffBasic coding and coding total length value;

When coding total length value during, but deduct basic coding length value value K, with j piece subimage difference coefficient DC with the total length value of encoding greater than the basic coding length value _DiffLeast significant bit be arranged in the binary number of K bit and subtract 1, the last binary number that obtains is for replenishing coding;

Basic coding and additional coded sequence are combined into j piece subimage difference coefficient DC _DiffThe coding of variable-length;

6. non zero AC coefficient in the one-dimension array of step in is 5. encoded one by one; JPEG coefficient coding classification chart among the Annex K is determined the AC class that non zero AC coefficient is affiliated one by one in the contrast Joint Photographic Experts Group;

According to the number of null value before aforementioned data that inquire the AC generic and the non zero AC coefficient, contrast basic coding and the coding total length value that the AC coefficient luminance component huffman coding table of giving tacit consent among the Joint Photographic Experts Group Annex K is inquired about each non zero AC coefficient with these two data;

When coding total length value during greater than the basic coding length value, but deduct basic coding length value value K with coding total length value, the least significant bit of AC coefficient is arranged in the binary number of K bit and subtracts 1, the last binary number that obtains is encoded for replenishing;

Basic coding and additional coded sequence are combined into the coding of the variable-length of non zero AC coefficient;

7. the coefficient that step is null value to the end item in the one-dimension array in is 5. represented with EOB sign EOB, the AC coefficient luminance component huffman coding table inquiry EOB that gives tacit consent among the contrast Joint Photographic Experts Group Annex K is encoded to 1010, and the coding of EOB is added into step coding end 6., finish the coding of j piece subimage in the i tomographic image;

8. according to step method 1.～7., (x, the coding of other subimage blocks y) all dispose up to all 576 number of sub images pieces to ask for the 1st tomographic image f;

Step (3) is compressed other 29 layers among the three-dimensional matrice img according to the method for step (2), finishes the lossy compression method of imaging data of multifocal multiphoton microscope.

Step (4) adopts image compression rate CR that the compression method of image is quantitatively described, and its expression formula is as follows: $C_R=\frac{n_1}{n_2}$

Wherein, n ₁Represent occupied storage size in the former data acquisition system (being original image), n ₂The data of representative after through coding in conjunction with (i.e. compression after image) occupied storage size.

Adopt use lossy compression method method compression multifocal multiphoton microscopic imaging video data block of the present invention, by using certain liver specimens data block file gfp_106000.int experiment to show, the compression ratio of the 1st tomographic image of this image block is 50.5852, the compression ratio of other 29 tomographic images is between 50～93, image after the compression descends to some extent than the original image quality, but can obtain considerable compression ratio.

Claims (1)

1. the lossy compression method method of an imaging data of multifocal multiphoton microscope is characterized in that this method comprises the steps:

Step (1) reads the binary format data file of multifocal multiphoton microscopic imaging, and generation resolution is that M * N, each pixel are 8, the image of L layer altogether; Wherein M, N and L are by the scanning imagery characteristic decision of multifocal multiphoton microscope; The image array that the binary format data file is formed the M * N resolution of L layer is stored among the three-dimensional matrice img;

Step (2) extract the i layer data of three-dimensional matrice img as the input f of compression (x, y), 1≤i≤L wherein, adopt jpeg algorithm to f (x y) compresses, and concrete steps are as follows:

1. (x y) is divided into nonoverlapping 8 * 8 subimage, and j piece subimage is designated as f with f ₈(x, y), wherein $1\≤j\≤\frac{M\×N}{8\×8};$

2. (x, y) each pixel of all subimages in from left to right deducts 2 from top to bottom to f ^K-1Do gray scale and move, obtain the subimage f ' after j piece subimage gray scale moves ₈(x, y), wherein k is the bit number of image;

3. the subimage f ' after j piece subimage gray scale being moved ₈(x y) carries out the Forward Discrete Cosine Transform conversion, obtains f ' ₈(x, Forward Discrete Cosine Transform T y) (u, v),

$T(u,v)=\underset{x=0}{\overset{7}{\mathrm{\Σ}}}\underset{y=0}{\overset{7}{\mathrm{\Σ}}}{f}_8^{\′}(x,y)\mathrm{\α}\left(u\right)\mathrm{\α}\left(v\right)\cos \left[\frac{(2x+1)\mathrm{u\π}}{16}\right]\cos \left[\frac{(2y+1)\mathrm{v\π}}{16}\right]$

$u=\mathrm{0,1,2},\·\·\·7,v=\mathrm{0,1,2},\·\·\·7,\mathrm{\α}\left(u\right)=\left\{\begin{array}{cc}\sqrt{\frac{1}{8}}& u=0\\ \frac{1}{2}& u=\mathrm{1,2},\·\·\·,7\end{array}\right.,$ $\mathrm{\α}\left(v\right)=\left\{\begin{array}{cc}\sqrt{\frac{1}{8}}& v=0\\ \frac{1}{2}& v=\mathrm{1,2},\·\·\·,7\end{array}\right.$

4. (u v) carries out normalization and quantification and obtains with T

$\hat{T}(u,v)=\mathrm{round}\left[\frac{T(u,v)}{Z(u,v)}\right],u=\mathrm{0,1},\·\·\·,7,v=\mathrm{0,1},\·\·\·,7$

In the formula, round rounds up to get the function of nearest integer, and (u v) is a luminance component quantization table in the Joint Photographic Experts Group Annex K item to Z;

5. will

Rearrange element position according to the Z-shaped pattern in the Joint Photographic Experts Group, obtain an one-dimension array that 64 elements are arranged, to first element DC coefficient in this one-dimension array

Carry out huffman coding with other elements A C coefficient that removes first element in this one-dimension array;

Ask for j piece subimage in the i tomographic image

And the DC difference coefficient DC between the j-1 piece subimage DC coefficient _Diff,

${\mathrm{DC}}_{\mathrm{diff}}={\hat{T}}_j\left(\mathrm{0,0}\right)-{\hat{T}}_{j-1}\left(\mathrm{0,0}\right),1\≤j\≤\left(\frac{M\×N}{8\×8}\right),{\hat{T}}_0\left(\mathrm{0,0}\right)=0$

JPEG coefficient coding classification chart among the Annex K is determined j piece subimage difference coefficient DC in the contrast Joint Photographic Experts Group _DiffAffiliated DC difference class, contrast the DC coefficient luminance component huffman coding table of giving tacit consent among the Joint Photographic Experts Group Annex K then and inquire about j piece subimage difference coefficient DC _DiffBasic coding and coding total length value;

When coding total length value during, but deduct basic coding length value value K, with j piece subimage difference coefficient DC with the total length value of encoding greater than the basic coding length value _DiffLeast significant bit be arranged in the binary number of K bit and subtract 1, obtain the additional coding of binary number at last;

Basic coding and additional coded sequence are combined into j piece subimage difference coefficient DC _DiffThe coding of variable-length;

6. non zero AC coefficient in the one-dimension array of step in is 5. encoded one by one; JPEG coefficient coding classification chart among the Annex K is determined the AC class that non zero AC coefficient is affiliated one by one in the contrast Joint Photographic Experts Group;

The number of null value before AC generic data that inquire and the non zero AC coefficient contrasts basic coding and the coding total length value that the AC coefficient luminance component huffman coding table of giving tacit consent among the Joint Photographic Experts Group Annex K is inquired about each non zero AC coefficient with these two data;

When coding total length value during, but deduct basic coding length value value K, the least significant bit of AC coefficient is arranged in the binary number of K bit and subtracts 1, obtain the additional coding of binary number at last with the total length value of encoding greater than the basic coding length value;

Basic coding and additional coded sequence are combined into the coding of the variable-length of non zero AC coefficient;

7. the coefficient that step is null value to the end item in the one-dimension array in is 5. represented with EOB sign EOB, the AC coefficient luminance component huffman coding table inquiry EOB that gives tacit consent among the contrast Joint Photographic Experts Group Annex K is encoded to 1010, and the coding of EOB is added into step coding end 6., finish the coding of j piece subimage in the i tomographic image;

8. according to step method 1.～7., ask for the coding of other subimage blocks of i tomographic image;

Step (3) is compressed other layers among the three-dimensional matrice img according to the method for step (2), finishes the lossy compression method of imaging data of multifocal multiphoton microscope.