CN106405546A - Quick correlated imaging system and method based on compression speckle - Google Patents

Abstract

The invention discloses a quick correlated imaging system and method based on compression speckle. Through compressed sampling technique, correlated imaging technique and sparse information recovery technique, compressed sampling is carried out on scene information by utilizing the compression speckle, and scene compression information is recovered by utilizing the compression speckle and detection intensity; through compression speckle, data volume is reduced, and imaging speed of the correlated imaging system is improved; and finally, recovery of complete and accurate scene information is realized through compressed sensing of the scene compression information.

Description

Quick relevance imaging system and method based on compression speckle

Technical field

The present invention relates to relevance imaging technical field, specifically a kind of quick relevance imaging system based on compression speckle and Method.

Background technology

Relevance imaging technology is studied by people always in recent years, and traditional association imaging system is carried out to scene using speckle light Irradiate, acquisition system is detected using single pixel and obtains scene echoes signal, using speckle light distribution information and detected intensity information Carry out related operation and obtain scene image information.Such system just develops towards practical implementation, but system needs different dissipating The collection of irradiation many times of speckle light could effectively obtain scene image information, and therefore system realtime imaging ability receives very about Bundle.At present, improving the image taking speed of relevance imaging system and quality is one of this area research direction.Natural using the overwhelming majority Scene image information have the characteristics that openness, this invention using compression speckle scene information is compressed sample, reduce Relevance imaging system data amount, can improve relevance imaging system operations speed, relevance imaging field have wide should Use prospect.

Content of the invention

It is an object of the invention to provide a kind of quick relevance imaging system and method based on compression speckle, to improve association The image taking speed of imaging system.

In order to achieve the above object, the technical solution adopted in the present invention is：

Based on compression speckle quick relevance imaging system and method it is characterised in that：Adjust including light source, beam expander, light Device DMD processed, projecting lens, plus lens, single pixel detector data acquisition system, the light that described light source sends is through beam expander After be irradiated on photomodulator DMD, photomodulator DMD light is modulated so that modulate light field generation speckle there is compression Sparse form, is capable of the compression sampling to scene, and the compression speckle that modulation light field produces irradiates scene, field through projecting lens Scape reflected light converges on single pixel detector through plus lens, and single pixel Detector and data acquisition system connects, by data Acquisition system obtains the total reflective light intensity information of scene by single pixel detector.

The described quick relevance imaging system and method based on compression speckle it is characterised in that：Photomodulator DMD is permissible Realize the modulation to light field, compression speckle can be produced, and this speckle distribution form accurately can be known by modulation intelligence, single Pixel detector can detect to total reflection intensity signal.

The described quick relevance imaging system and method based on compression speckle it is characterised in that：Using compression speckle pair Object information is compressed the data volume significantly reducing in relevance imaging of sampling, and realizes quick relevance imaging.

The quick relevance imaging method of described system it is characterised in that：Comprise the following steps：

(1), fix condensation matrix first, then carry out different sparse samplings successively on the basis of condensation matrix and add It is downloaded to realize on photomodulator DMD compressing speckle, using compression speckle, scene is irradiated, single pixel detector is to scene Reflected signal is detected；Acquisition system is acquired and protects to the reflected signal that much individual different compression speckle irradiates scene Deposit；Setting condensation matrix B, its be M × M size containing 0,1 matrix, wherein the points of efficiently sampling 1 be N × N, meet condition M>N, data compression rate γ=N²/M²；Assume that jth time utilizes sparse sampling matrix S_jCompression after being acted on condensation matrix B Speckle is irradiated to scene, and the intensity signal that acquisition system obtains is i, then this process can be expressed as shown in formula (1)：

I=∑_M×MBS_jR, (1),

Here, R represents scene reflectivity rate information, ∑_M×MRepresent that the element to M × Metzler matrix is sued for peace；

(2), using the efficiently sampling location index in condensation matrix, adjustment of matrix is carried out to the speckle irradiating scene, reject Non- sampled point in data is 0 value point in condensation matrix, and data volume substantially reduces, and can effectively improve the computing speed of relevance imaging Degree, can obtain compression scene information after entering line algorithm computing；Using condensation matrix B, speckle is compressed, only select and adopt Sampling point is 1 position, according to matrix N × N, above formula (1) is rewritten as shown in formula (2)：

I=∑_N×NS_j' R ', (2),

Wherein, S_j' it is BS_jThrough rearranging form, R ' is that the sampled point of R rearranges form, and size is N × N, that is, Compressed format for prime information；According to formula (2) form to k₁The process of secondary speckle illumination collection is expressed in matrix as formula (3) shown in：

Wherein,For k₁The intensity signal form of secondary collection, size is k₁× 1,Column vector representation for R ', For k₁The matrix representation forms of secondary different compression speckle, size is k₁×N²；Need to locate by can be seen that to the analysis of formula The data matrix size of reason is N²×k₁, the compression ratio of data is γ；Meanwhile due to needing to solve the number phase of R ' unknown number A lot of for will lacking of R, therefore can reduce speckle and irradiate number of times, in real system, the compression degree of data volume also will than γ Little；

(3), utilize above-mentioned steps obtain compression scene information, according to efficiently sampling location index carry out arrangement obtain with Former scene sparse information in the same size, can obtain the scene information of complete and accurate using sparse information recovery technique；According to pressure Effective index information handle of contracting matrixIt is adjusted to the sparse scene information R that size is M × M_B；Using compressed sensing sparse information Recovery technique, by R_BRestore the scene information R of complete and accurate, solving optimum formula is：

argmin[||R_B-BR||²+λΦ_TV(R)] (4),

In formula (4), Φ_TVRepresent total variation model function, λ represents smoothing factor.

In the present invention, photomodulator DMD is modulated to light according to compression speckle form, produces compression speckle, is capable of Compression sampling to scene.Realize the fully sampled different of scene information from traditional association imaging using speckle, present invention achieves The compression sampling of scene information, data volume significantly reduces, and can improve the image taking speed of relevance imaging.

It is an advantage of the invention that：The present invention combines compression sampling technology, relevance imaging technology and sparse information recovery technique, Reduce data volume using compression speckle, realize, using information correlativity, the quick obtaining that scene compresses information, improve relevance imaging The image taking speed of system, realizes the recovery of complete and accurate scene information finally by compressed sensing, and the method is led in relevance imaging Domain is with a wide range of applications.

Brief description

Fig. 1 is the structural representation of the present invention.

Fig. 2 system experimentation result figure, wherein：

Fig. 2 a is the subject image being embodied as will be imaged in experiment, and Fig. 2 b is using condensation matrix acquisition of information Object sparse information figure, Fig. 2 c is using compressed sensing, sparse information to be carried out restoring the hum pattern obtaining.

Specific embodiment

As shown in figure 1, based on the quick relevance imaging system compressing speckle, its device includes light source 1, beam expander 2, light tune Device DMD3 processed, projecting lens 4, plus lens 5, single pixel detector 6 data acquisition processing system 7；

Light source 1 lights, and has larger cross-sectional distribution, then light beam is irradiated on photomodulator DMD3 after beam expander 2 Produce compression speckle through projecting lens 4, scene to be irradiated, scene reflectivity light reaches single pixel detector 6 through plus lens 5 On, then carry out data acquisition, preservation and process through data acquisition processing system 7.

Fix condensation matrix first, then carry out different sparse samplings successively on the basis of condensation matrix to realize pressing Contract sparse speckle, using compressing sparse speckle, scene is irradiated, single pixel detector detects to scene reflectivity signal. Acquisition system is acquired and preserves to the different many times reflected signals compressing sparse speckle irradiation scene.Setting condensation matrix B, its be M × M size containing 0,1 matrix, wherein the points of efficiently sampling 1 be N × N, meet condition (M>N), data compression Rate γ=N²/M².Assume that jth time utilizes sparse sampling matrix S_jCompression speckle after being acted on condensation matrix B is entered to scene Row irradiates, and the intensity signal that acquisition system obtains is i, then this process can be expressed as：

I=∑_M×MBS_jR, (1),

Here, R represents scene reflectivity rate information, ∑_M×MRepresent that the element to M × Metzler matrix is sued for peace.

Using the efficiently sampling location index in condensation matrix, adjustment of matrix is carried out to the associated compression speckle irradiating scene, The non-sampled point rejected in data is 0 value point in condensation matrix, and data volume substantially reduces, and can effectively improve the fortune of relevance imaging Calculate speed, compression scene information after entering line algorithm computing, can be obtained.Using condensation matrix B, speckle is compressed, only selected Going out sampled point is 1 position, according to matrix N × N, above formula (1) is rewritten as：

I=∑_N×NS_j' R ', (2),

Wherein, S_j' it is BS_jThrough rearranging form, R ' is that the sampled point of R rearranges form, and size is N × N, that is, Compressed format for prime information.According to formula (2) form to k₁The process of secondary speckle illumination collection is expressed in matrix as：

Wherein,For k₁The intensity signal form of secondary collection, size is k₁× 1,Column vector representation for R ', For k₁The secondary different matrix representation forms compressing sparse speckle, size is k₁×N².Adopted by can be seen that to the analysis of formula Carried out with traditional method during Algorithm for Solving, needing data matrix size to be processed to be M²×k₁, and need to process using this patent Data matrix size be N²×k₁, the compression ratio of data is γ.Meanwhile due to needing the number solving R ' unknown number relatively A lot of will lacking of R, therefore can reduce speckle and irradiate number of times, in real system, the compression degree of data volume is also less than γ.

The compression scene information being obtained using above-mentioned steps, is carried out arrangement according to efficiently sampling location index and obtains and former field Scape sparse information in the same size, can obtain the scene information of complete and accurate using sparse information recovery technique.According to compression square Effective index information handle of battle arrayIt is adjusted to the sparse scene information R that size is M × M_B.Restored using compressed sensing sparse information Technology, by R_BRestore the scene information R of complete and accurate, solving optimum formula is：

argmin[||R_B-BR||²+λΦ_TV(R)] (4),

Here, Φ_TVRepresent total variation (TV) pattern function, λ represents smoothing factor.

Obviously, those skilled in the art can be by carrying out to compression speckle relevance imaging system involved in the present invention Change with modification without departing from the spirit and scope of the present invention.So, want if these modifications and variation belong to right of the present invention Ask and its equivalent technologies within the scope of, then the present invention is also intended to comprise these modifications and modification.

Experimental result

In order to verify the feasibility of the present invention, design system is simultaneously once tested, and experimental result is as shown in Figure 2.Real As shown in Figure 2 a, imaging resolution size is 80 × 80 to the object will being imaged in testing.Using condensation matrix efficiently sampling number For 4096, that is, compression ratio is 0.64.Using compressing sparse speckle, object is carried out with 3400 irradiations, obtained using condensation matrix information The object sparse information taking is as shown in Figure 2 b.Finally sparse information is carried out restore the information obtaining such as Fig. 2 c using compressed sensing Shown, by result it can be seen that the method can effectively obtain object information.

Claims (4)

1. based on compression speckle quick relevance imaging system and method it is characterised in that：Including light source, beam expander, light modulation Device DMD, projecting lens, plus lens, single pixel detector data acquisition system, the light that described light source sends is after beam expander Be irradiated on photomodulator DMD, photomodulator DMD light is modulated so that modulate light field generation speckle have compression dilute Thin form, is capable of the compression sampling to scene, and the compression speckle that modulation light field produces irradiates scene, scene through projecting lens Reflected light converges on single pixel detector through plus lens, and single pixel Detector and data acquisition system connects, and is adopted by data Collecting system obtains the total reflective light intensity information of scene by single pixel detector.

2. according to claim 1 based on compression speckle quick relevance imaging system and method it is characterised in that：Light is adjusted Device DMD processed can realize the modulation to light field, can produce compression speckle, and this speckle distribution form can pass through modulation intelligence Accurately know, single pixel detector can detect to total reflection intensity signal.

3. according to claim 1 based on compression speckle quick relevance imaging system and method it is characterised in that：Using Compression speckle is compressed the data volume significantly reducing in relevance imaging of sampling to object information, realizes quickly being associated to Picture.

4. the quick relevance imaging method based on system described in claim 1 it is characterised in that：Comprise the following steps：

(1), fix condensation matrix first, then carry out different sparse samplings successively on the basis of condensation matrix and be loaded into To realize on photomodulator DMD compressing speckle, using compression speckle, scene to be irradiated, single pixel detector is to scene reflectivity Signal is detected；Acquisition system is acquired and preserves to the reflected signal that much individual different compression speckle irradiates scene；If Put condensation matrix B, its be M × M size containing 0,1 matrix, wherein the points of efficiently sampling 1 be N × N, meet condition M>N, Data compression rate γ=N²/M²；Assume that jth time utilizes sparse sampling matrix S_jCompression speckle after being acted on condensation matrix B Scene is irradiated, the intensity signal that acquisition system obtains is i, then this process can be expressed as shown in formula (1)：

I=Σ_M×MBS_jR, (1),

Here, R represents scene reflectivity rate information, ∑_M×MRepresent that the element to M × Metzler matrix is sued for peace；

(2), using the efficiently sampling location index in condensation matrix, adjustment of matrix is carried out to the speckle irradiating scene, reject data In non-sampled point be 0 value point in condensation matrix, data volume substantially reduces, and can effectively improve the arithmetic speed of relevance imaging, Compression scene information can be obtained after entering line algorithm computing；Using condensation matrix B, speckle is compressed, only selected sampled point For 1 position, according to matrix N × N, above formula (1) is rewritten as shown in formula (2)：

I=0_N×NS′_jR ', (2),

Wherein, S '_jFor BS_jThrough rearranging form, R ' is that the sampled point of R rearranges form, and size is N × N, as former The compressed format of information；According to formula (2) form to k₁The process of secondary speckle illumination collection is expressed in matrix as formula (3) institute Show：

$\stackrel{\‾}{I}={\stackrel{\‾}{S}}^{\′}{\stackrel{\‾}{R}}^{\′},---\left(3\right),$

Wherein,For k₁The intensity signal form of secondary collection, size is k₁× 1,Column vector representation for R ',For k₁Secondary The matrix representation forms of different compression speckles, size is k₁×N²；Need number to be processed by can be seen that to the analysis of formula It is N according to matrix size²×k₁, the compression ratio of data is γ；Meanwhile due to needing the number solving R ' unknown number with respect to R To lack a lot, therefore can reduce speckle and irradiate number of times, in real system, the compression degree of data volume is also less than γ；

(3), utilize the compression scene information that above-mentioned steps obtain, carry out arrangement according to efficiently sampling location index and obtain and former field Scape sparse information in the same size, can obtain the scene information of complete and accurate using sparse information recovery technique；According to compression square Effective index information handle of battle arrayIt is adjusted to the sparse scene information R that size is M × M_B；Restored using compressed sensing sparse information Technology, by R_BRestore the scene information R of complete and accurate, solving optimum formula is：

argmin[||R_B-BR||²+λΦ_TV(R)] (4),

In formula (4), Φ_TVRepresent total variation model function, λ represents smoothing factor.