CN106228169A - The distance extracting method in holoscan space based on discrete cosine transform - Google Patents

Abstract

The invention discloses a kind of holoscan space length extracting method based on discrete cosine transform, belong to optical holographic scanning field.The principle of the extraction distance that the present invention provides is: carries out discrete cosine transform to rebuilding image, thus extracts the sharpness information rebuilding image.It mainly solves the problem of axial location of Automatic-searching object under test.The present invention can efficiently solve the problem of object space detection, and it is applicable to multiple field.

Description

The distance extracting method in holoscan space based on discrete cosine transform

Technical field

The invention belongs to optical holographic scanning field, relate to a kind of holoscan space length based on discrete cosine transform Extracting method.

Background technology

Optical scanning holographic technique, is called for short OSH, is a branch of Digital Holography, and it is by Poon and Korpel Proposing the earliest, the information of 3-dimensional object can be stored as the hologram of 2 dimensions by OSH.Since this technology proposes, in scanning It is used widely in the fields such as holographic microscope, 3D rendering identification and 3D optical remote sensing.

Space length extractive technique is an important technology in OSH, and it can extract the axial location of object under test Information, and the axial position information of object under test is by the important parameter of image reconstruction.Therefore, space length extractive technique is also It is provided with important researching value.Research proves, in interative computation, can find out reconstruction image by discrete cosine transform sharp The local maximum of degree, when obtaining the local maximum of acutance, can deduce the axial position information of object under test, thus Extract the space length of object.

Document " Three-dimensional matching by use of phase-only holographic Information and the Wigner distribution " propose utilize Wigner distribution extract object under test Axial space information, but the method may be only available for two objects with dependency, itself has limitation, is difficult to push away Extensively.

Document " Blind sectional image reconstruction for optical scanning Holography " propose to utilize Prewitt operator extraction to rebuild image border, this method has preferable practicality, but Due to the existence of out of focus noise edge, have impact on the degree of accuracy of result of calculation.

Summary of the invention

The invention provides a kind of holoscan space length extracting method based on discrete cosine transform, the method utilizes Discrete cosine transform, extracts image low-frequency component, removes radio-frequency component, thus extracts the acutance of image.

The technical solution used in the present invention is:

A kind of holoscan space length extracting method based on discrete cosine transform, its flow process as it is shown in figure 1, include with Lower step:

Laser is divided into two-beam road by first polarization beam apparatus by step 1., and wherein a light beam passes sequentially through one Acousto-optic modulator and beam expander, after another light beam passes sequentially through another acousto-optic modulator, beam expander and convex lens, by this two Bar light beam interferes formation Fresnel interference fringe after second polarization beam apparatus optically focused；

Step 2. utilizes Fresnel interference fringe measuring targets to be scanned, and by photoelectric detector receive scanning after Optical information, thus obtain the hologram of object under test；

Step 3. arranges initial value and the higher limit thereof of distance parameter, and the scanning step of setpoint distance parameter, with described initial value Currency as distance parameter；

After the hologram of gained is done Fourier transformation by step 4., transmit function with the traditional optical with distance parameter After the conjugate multiplication of frequency-domain expression, then through inverse Fourier transform, i.e. can get the reconstruction image of object under test；

Step 5. is divided into the subgraph of n*n by rebuilding image, and subgraph carries out the discrete cosine transform of n*n；Calculate, remember Record the sharpness value of picture through discrete cosine transform and the distance parameter value of correspondence thereof；

The currency of distance parameter is added the scanning step last look as distance parameter by step 6., and repeats step 4 To step 5 when the currency of distance parameter reaches set higher limit, be derived from multiple sharpness value, Qi Zhongju The distance parameter value that portion's maximum is corresponding is object under test location on axial space.

The invention has the beneficial effects as follows:

(1) principle extracting object under test axial distance that the present invention provides, utilizes dct algorithm to realize right The axial space information retrieval of object under test, holoscan space length extractive technique range of application includes that biological cell is observed, Optical focus etc., therefore the present invention is applicable to every field, and range of application is the widest.

(2) present invention first becomes the subgraph of 8*8 to the picture segmentation rebuild, and utilizes discrete cosine transform to become subgraph Change, calculate its sharpness value thus realize the extraction of the axial location of object.

(3) present invention not only implementation is simple, convenient to operate, has the strongest practicality simultaneously, is suitable for promoting the use of.

Accompanying drawing explanation

The method flow schematic diagram that Fig. 1 provides for the present invention；

Fig. 2 is the basic block diagram of the embodiment of the present invention；

Fig. 3 is the object under test schematic diagram of the embodiment of the present invention；

Fig. 4 is the object under test hologram of the embodiment of the present invention；

Fig. 5 is the reconstruction image of the embodiment of the present invention；

Fig. 6 is the discrete cosine transform design sketch to reconstruction image of the embodiment of the present invention；

Fig. 7 is the acutance functional arrangement of the embodiment of the present invention.

Detailed description of the invention

This detailed description of the invention adopts the following technical scheme that

A kind of holoscan locus based on discrete cosine transform extracting method, its flow process as it is shown in figure 1, include with Lower step:

Laser is divided into two-beam road by first polarization beam apparatus by step 1., and wherein a light beam passes sequentially through one Acousto-optic modulator and beam expander, after another light beam passes sequentially through another acousto-optic modulator, beam expander and convex lens, by this two Bar light beam interferes formation Fresnel interference fringe after second polarization beam apparatus optically focused；

The system structure that this example uses is as in figure 2 it is shown, the light that angular frequency is ω sent by same light source is by first Polarization beam apparatus BS1 is divided into two bundles, the most a branch of by acousto-optic modulator AOFS1 and convex lens L1, forms spherical wave；Another bundle Through acousto-optic modulator AOFS2, form plane wave；Two-beam is interfered after second polarization beam apparatus polymerization before object under test Forming Fresnel interference fringe, its expression formula on axial space residing for object under test (as shown in Figure 3) is as follows:

$h(x,y;z_i)=\frac{-j}{{\mathrm{\λz}}_i}\exp \[\frac{j\mathrm{\π}}{{\mathrm{\λz}}_i}(x^2+y^2)\]---\left(1\right)$

Wherein, object under test is divided into n-layer, z in the axial direction_iRepresent the axial location at i-th layer of place of object under test, λ Representing optical maser wavelength, x and y represents the lateral coordinates of object；

Step 2. is scanned by above-mentioned Fresnel interference fringe measuring targets, and utilizes photoelectric detector to receive Optical information after scanning, thus obtain the hologram of object under test；

This example utilizes Fresnel interference fringe to be scanned object, and hologram (such as Fig. 4) storage formed is arrived meter In calculation machine, specific as follows:

$g(x,y)=\underset{i=1}{\overset{i=n}{\Σ}}{F}^{-1}\left\{F\right\{I(x,y;z_i)\}\×F\left(h\right(x,y,z_i\left)\right)\}---\left(2\right)$

Wherein, (x y) represents hologram, F to g^-1, F represent inverse Fourier transform and Fourier transformation respectively, i represents to be measured Object in the axial direction i-th layer, I (x, y；z_i) represent determinand complex amplitude function；

Step 3. arranges initial value and the higher limit thereof of distance parameter, and the scanning step of setpoint distance parameter, with described initial value Currency as distance parameter；In this example, arranging distance parameter initial value is z_o=600mm, higher limit is z_t=1100mm；

After the hologram of gained is done Fourier transformation by step 4., transmit letter with the traditional optical with described distance parameter Frequency-domain expression (the F of number^*{h(x,y；z_j)) conjugate multiplication after, then through inverse Fourier transform, i.e. can get object under test Reconstruction image, as shown in Figure 5；

In this example, carry out embodying of image reconstruction by optical transfer function as follows:

$\begin{array}{c}{I}_j(x,y)=F^{-1}\left\{F\right\{g(x,y)\}\×F^{*}\{h(x,y;z_j)\left\}\right\}\\ =I(x,y;z_j)+\underset{i\≠j}{\Σ}I(x,y;z_i)\⊗h(x,y;z_j)\end{array}---\left(3\right)$

Wherein, I_j(x y) represents the reconstruction image information of object under test jth layer, z_jRepresent that object under test rebuilds image jth The axial location at layer place, * represents conjugation,Represent convolution；

Step 5. is divided into the subgraph of n*n by rebuilding image, and subgraph carries out the discrete cosine transform of n*n, calculates, remembers Record the sharpness value of picture (as shown in Figure 6) through discrete cosine transform and the distance parameter value of correspondence thereof.Continuous increasing along with n Greatly, the more useful information of object can be extracted, but when exceeding certain limit, the useful information of extraction is not significantly increased, but Add the complexity of calculating, so selecting n=8 in this example；

In this example, reconstruction image being divided into the subgraph of 8*8, expression is as follows:

$S_{ij}^8(x,y)={\left\{I_j(x,y)\right\}}_{x=8i,y=8j}^{x=8i+7,y=8j+7}---\left(4\right)$

WhereinRepresent the subgraph rebuilding image.Utilize discrete cosine transform to rebuilding at the subgraph of image The expression of reason is as follows:

$F(u,v)=c\left(u\right)c\left(v\right)\underset{x=0}{\overset{M}{\Σ}}\underset{y=0}{\overset{N}{\Σ}}{S}_{ij}^8(x,y)\cos \frac{\mathrm{\π}(2x+1)u}{2M}\cos \frac{\mathrm{\π}(2y+1)v}{2N}$

Wherein (u v) represents the discrete cosine transform of reconstruction image to F；M*N represents the size rebuilding picture；C (u), c (v) Represent as follows:

$\begin{array}{cc}c\left(u\right)=\sqrt{1/M}& u=0\\ c\left(u\right)=\sqrt{2/M}& u=1,\mathrm{2...},M-1\end{array}---\left(5\right)$

$\begin{array}{cc}c\left(v\right)=\sqrt{2/N}& v=0\\ c\left(v\right)=\sqrt{2/N}& v=1,\mathrm{2...},N-1\end{array}---\left(6\right)$

Calculating the sharpness value rebuilding image, its expression is as follows:

$M=\frac{\underset{i=j=0}{\overset{M/8}{\Σ}}\underset{u+v\≤8i+8j+6}{\overset{u+v\≥8i+8j}{\Σ}}|F_C(u,v){|}^2}{{\left(\underset{i=j=0}{\overset{M/8}{\Σ}}\underset{u+v\≤8i+8j+6}{\overset{u+v\≥8i+8j}{\Σ}}\right|F_C(u,v)\left|\right)}^2}---\left(7\right)$

Wherein F_C(u is v) that (6 is to extract through discrete cosine transform image low-frequency component to F for u, normalization matrix v) Threshold value, wherein 6 is the empirical value gone out by experimental calculation.

The currency of distance parameter is added the scanning step last look as distance parameter by step 6., and repeats step 4 To step 5 when the currency of distance parameter is equal to higher limit, thus can obtain multiple sharpness value, two of which local The distance parameter value (as shown in Figure 7) that maximum is corresponding is object under test location on axial space.

Claims (5)

1. a holoscan space length extracting method based on discrete cosine transform, comprises the following steps:

Laser is divided into two-beam road by first polarization beam apparatus by step 1., and wherein a light beam passes sequentially through an acousto-optic Manipulator and beam expander, after another light beam passes sequentially through another acousto-optic modulator, beam expander and convex lens, by these two light Bundle interferes formation Fresnel interference fringe after second polarization beam apparatus optically focused；

Step 2. utilizes Fresnel interference fringe measuring targets to carry out optical holographic scanning, and is swept by photoelectric detector reception Optical information after retouching, it is thus achieved that the hologram of object under test；

Step 3. arranges initial value and the higher limit thereof of distance parameter, the scanning step of setpoint distance parameter, using described initial value as The currency of distance parameter；

After the hologram of gained is done Fourier transformation by step 4., with the frequency domain that the traditional optical with distance parameter transmits function After the conjugate multiplication of expression formula, then through inverse Fourier transform, obtain the reconstruction image of object under test；

Step 5. is divided into the subgraph of n*n by rebuilding image, and subgraph carries out the discrete cosine transform of n*n；Calculate, record warp Cross the sharpness value of the picture of discrete cosine transform and the distance parameter value of correspondence thereof；

The currency of distance parameter is added the scanning step last look as distance parameter by step 6., and repeats step 4 to step Till rapid 5 when the currency of distance parameter reaches set higher limit, being derived from multiple sharpness value, wherein local is The distance parameter value of big value correspondence is the axial location of object under test.

Holoscan space length extracting method based on discrete cosine transform the most according to claim 1, its feature exists In: Fresnel interference fringe described in step 1 is as follows in the expression formula of axial positions residing for object under test:

$h(x,y;z_i)=\frac{-j}{{\mathrm{\λz}}_i}\exp \[\frac{j\mathrm{\π}}{{\mathrm{\λz}}_i}(x^2+y^2)\]---\left(1\right)$

Wherein, object under test is divided into n-layer, z in the axial direction_iRepresenting the axial location at i-th layer of place of object under test, λ represents sharp Optical wavelength, x and y represents the space coordinates of object.

Holoscan space length extracting method based on discrete cosine transform the most according to claim 2, its feature exists In: the hologram information described in step 2 is specific as follows:

$g(x,y)=\underset{i=1}{\overset{i=n}{\Σ}}{F}^{-1}\left\{F\right\{I(x,y;z_i)\}\×F\left(h\right(x,y,z_i\left)\right)\}---\left(2\right)$

Wherein, (x y) represents hologram, F to g^-1, F represent inverse Fourier transform and Fourier transformation respectively, i represents object under test I-th layer in the axial direction, I (x, y；z_i) represent determinand complex amplitude function.

Holoscan space length extracting method based on discrete cosine transform the most according to claim 3, its feature exists In, embodying of the reconstruction image described in step 4 is as follows:

$\begin{array}{c}{I}_j\left(x,y\right)=F^{-1}\left\{F\left\{g\left(x,y\right)\right\}\×F^{*}\left\{h\left(x,y;z_j\right)\right\}\right\}\\ =I\left(x,y;z_j\right)+\underset{i\≠j}{\mathrm{\Σ}}I\left(x,y;z_i\right)\⊗h\left(x,y;z_j\right)\end{array}---\left(3\right)$

Wherein, I_j(x y) represents the reconstruction image information of object under test jth layer, z_iRepresent the axle at i-th layer of place of object under test To position, z_jRepresenting that object under test rebuilds the axial location at image jth layer place, * represents conjugation,Represent convolution.

5. according to the holoscan space length extraction side based on discrete cosine transform according to any one of claim 1-4 Method, it is characterised in that: in step 5, n value is 8.