CN102663376A - Near-infrared multi-intensity finger vein image acquisition and integration system, and method - Google Patents

Abstract

The invention discloses a near-infrared multi-intensity finger vein image acquisition and integration system, comprising a near-infrared multi-intensity acquisition terminal of finger vein images and a PC image processing terminal. A multi-intensity near-infrared light is utilized in the invention to image finger veins by transmission, which provides a significant improvement compared to multi-exposure image acquisition. A light source is regulated after an active analysis of gray distribution of the acquired image, wherein the light source has a high intensity level and is able to achieve a non-liner light intensity distribution, thereby providing a good controllability and high responding speed, and broadening research and application methods of near-infrared light source. A pixel-level fusion is carried out after an acquisition of a near-infrared multi-intensity finger vein image sequence, and the final fusion image has a high dynamic range, a high signal-to-noise ratio, uniform contrast and gray level, thereby improving the quality of the figure vein image.

Description

Obtaining and emerging system and method for many light intensity of near infrared finger venous image

Technical field

The present invention relates to personal identification and distinguishment technical field, the obtaining and emerging system and method for in particular a kind of many light intensity of near infrared finger venous image.

Background technology

Current vein identification technology is that use in the forward position of living things feature recognition in the body; Compare with fingerprint identification technology; Use finger vein features to carry out person identification following advantage is arranged: characteristic in the subepidermal body, daily life can not stay trace, is difficult to steal and forge; Vivo identification has only the blood that proper flow is arranged in the vein, just can collect vein image, can not stolen by violence; Noncontact is gathered, and does not have finger health and the stained problem of sensor, uses comfortable; Safe class is high, and the similar probability of veinprint is much smaller than fingerprint.

In view of above-mentioned advantage, finger vein identification technology has been applied to following field: financial terminal, and 2008, the bank ATM machine of Japan existing 80% adopted the authentication mode of vein identification; Computer network, Fujitsu has released the computer log terminal device based on palm vein; The place gate inhibition of guild hall, in the Beijing Olympic match, the women's softball game venue has successfully been used the hand back vein gate inhibition as security system; Means of transportation and instrument, the Singapore metro ticket checking realizes finger vena identification, HIT develops car door and the bearing circle that embeds vein identification; The intelligent building system, in November, 2008, " referring to vein identification Work attendance management system " formally comes into operation in Shanghai World Financial Center.

But because finger vena is concealed under epidermis; When existing near-infrared transmission finger carries out the phlebography imaging; Because the physiological tissue of finger is inhomogeneous bigger with the infrared imaging noise; Exist fuzzy, low, the uneven problem of gray scale of contrast of veinprint in individual finger venous image that obtains, the present invention proposes to want combining adaptive illumination to control the quality that (hardware) and image co-registration (software) improve the finger venous image that collects in finger vena research.

When adaptive illumination control mechanism mainly is transmission of near infra red light finger vena contrast imaging; Can regulate the near infrared intensity of illumination according to physiological tissue's thickness distribution of finger; Light intensity control just can adaptively be adjusted through simple gradation of image analysis, can collect many finger venous images that light intensity is more suitable fast through this method.

Image co-registration mechanism is to utilize the redundant information between many light intensity of near infrared finger venous image; Can veinprint be formed images preferably in partial fusion to an image; Merge the dynamic range that can improve finger venous image, enhancing contrast ratio, and homogenising gray scale through Pixel-level.

Summary of the invention

Technical matters to be solved by this invention is that the deficiency that is directed against prior art provides obtaining and emerging system and method for a kind of many light intensity of near infrared finger venous image.

Technical scheme of the present invention is following:

Obtaining and emerging system of a kind of many light intensity of near infrared finger venous image comprises the many intensity collections terminal and the PC image processing terminal of near infrared finger venous image; Wherein many intensity collections terminal of near infrared finger venous image comprises: led array light source, LED driving, microcontroller, near infrared camera; Said led array light source is used to provide the near infrared light of transmission finger; The LED driver module is that each LED in the led array light source provides independently constant current source driving; Micro controller module is used to control the LED driving provides drive signal to the led array light source; Control the near infrared camera simultaneously and catch the image of transmission of near infra red light finger vena radiography, micro controller module also is used for communicating with view data with PC and exchanges; The PC image processing terminal is used for the AP that the many intensity collections terminal to the near infrared finger venous image collects and merges.

Obtaining and fusion method of a kind of many light intensity of near infrared finger venous image comprises many light intensity finger vein image acquisition step and finger venous image fusion steps; Many light intensity finger vein image acquisition step comprises: A1, LED driver module send one group of tentative light intensity control signal; A2, microcontroller obtain vein image this moment through the near infrared camera; A3, microcontroller carry out image quality analysis to this vein image; Be equally divided into several row pieces to image by the pixel count that is listed as; Calculate the average gray value of each row piece; If the quantity of the row piece within suitable gray-scale value is not greater than 2 for average gray value, then looking this image is invalid image, needs to revise the light intensity control signal of corresponding row piece; If the row number of blocks of average gray value in suitable gray-scale value is more than or equal to 8, then looking this image is AP; A4, judge whether to gather the AP of sufficient amount, if " denying " gets into A5, if " be ", get into A6; A5, LED driver module send new light intensity control signal, and repeating step A2-A4 is up to the AP that obtains sufficient amount; A6, the collection of many light intensity finger venous image sequence are accomplished;

The finger venous image fusion steps comprises: B1, image block, said AP is all carried out piecemeal, and the piecemeal rule is to be equally divided into 10 row pieces by the col width pixel count, and calculates the average gray value of each row piece; B2, image column piece quality discrimination are promptly differentiated the image quality of each image column piece; B3, weights distribute and also carry out the level and smooth of image column interblock, differentiate according to the image quality of image column piece, and the image column piece of various piece is carried out the weighting weight allocation, obtain the concrete apportioning cost of weights of each image column piece; B4, image column piece merge, and according to the concrete apportioning cost of weights, each image column piece are carried out weighting fusion; B5, syncretizing effect are differentiated; B6, fusion finish.

Described method, said step B3 specifically carries out following steps: B31, supposes to have N to open many light intensity of near infrared finger venous image, is designated as I _n(n=1,2 ..., N), every image is divided into P image column piece, N*P image column piece so just arranged, note is I _i(i=1,2 ..., N*P), the average gray of each image column piece note is B _i, divide the informative weight value coefficient S of this image column piece of timing at weights _iCalculate by following formula (I):

S _i(x，y)＝exp[α·h(B _i)] (I)

(x, y) the row and column coordinate of remarked pixel in image, h (B in this formula _i) representing the ordinate value of average gray value Bi on Fig. 5 (c) curve of this image column piece, α is a smoothing factor;

B32, carry out the level and smooth of image column interblock, carry out the space smoothing coefficient G that weights distribute by distance _i, calculate by following formula (II):

$G_i(x,y)=\exp [-\frac{{(y-y_i)}^2}{2{\mathrm{\σ}}^2}]---\left(\mathrm{II}\right)$

In formula (II), σ is the standard deviation of Gaussian function, y _iBe i open under the illumination condition selected take out the central series coordinate of optimized image row piece; If a plurality of optimized image row pieces are arranged; The central series coordinate of the nearest best pixel piece of chosen distance current pixel piece then is not if all images row piece under this illumination condition is optimized image row piece, y so _iGet 0; Can confirm the space smoothing coefficient G of each pixel behind many light intensity image sequence piecemeal by this formula _i

B33, the final associating weights partition factor of calculating, see formula (III):

w _i(x，y)＝S _i(x，y)G _i(x，y) (III)

In formula (III), w _iBe the weights partition factor of associating, can this coefficient be carried out normalization, be designated as

Promptly obtained the concrete apportioning cost of weights of each image column piece.

The method of many light intensity near infrared finger vein image acquisition that the present invention proposes; Transmission of near infra red light finger vena contrast imaging method to existing single light intensity has been carried out significant improvement; Improved the automatic adjust intensity ability of acquisition system, differences of Physiological that can be individual according to finger is like dynamic light intensity that changes light source such as finger thickness, skin roughness; Innovated the acquisition methods of finger venous image, to the vein image of other body surfaces obtain and enhancing has reference function.

The present invention uses the transmission of near infra red light finger vena contrast imaging of multiple light intensity; Collection with respect to many exposure images has bigger improvement, and the gradation of image that active analysis of the present invention collects is adjusted light source after distributing, and the intensity progression of light source is high; Can realize that non-linear light intensity distributes; Controllability is good, and response speed is fast, has expanded the development and methods for using them of near-infrared light source.

The feasible method of the near infrared finger venous image sequence fusion of many light intensity is carried out in the present invention's proposition; Image block method based on the mean pixel intensity profile is proposed aspect the Pixel-level fusion; In the method that has proposed aspect the image quality differentiation of near-infrared image based on the quantitative test of optical sensor illumination response curve; Divide the weight of the image column piece that timing takes into full account good imaging quality and the transition weight of adjacent image row piece merging weights; Merge and contrast height, intensity profile is even, intensity profile is level and smooth, signal to noise ratio (S/N ratio) is high finger venous image; Expanded the weighting fusion method of many exposure images, innovated to some extent aspect the collection of near infrared finger venous image and the enhancing.

Description of drawings

Fig. 1 is the design proposal block diagram of many light intensity near infrared finger vein image acquisition and fusion;

Fig. 2 is the collection of many light intensity finger venous image sequence;

Fig. 3 be for the first time, for the second time, for the third time, finger venous image and the piecemeal average gray thereof gathered for the 4th time, numbering is respectively (a) and (b), (c), (d);

Fig. 4 is many light intensity finger venous image fusion process;

Fig. 5 is the illumination response function curve f of optical sensor, is numbered (a), and the inverse function curve g of f is numbered (b), and the derivative curve h of g is numbered (c);

Fig. 6 is the optimized image row piece that selects;

Fig. 7 distributes weights for the image column piece by illumination response function curve corresponding formula (I);

Fig. 8 distributes weights for the distance of optimized image row piece under image column piece and the identical illumination condition;

Fig. 9 distributes weights by Fig. 7 and comprehensive each image column piece of Fig. 8;

Figure 10 is the final width of cloth finger venous image that obtains that merges.

Embodiment

Below in conjunction with specific embodiment, the present invention is elaborated.

Embodiment 1

With reference to figure 1, obtaining and emerging system of many light intensity of near infrared finger venous image comprises the many intensity collections terminal and the PC image processing terminal of near infrared finger venous image; Wherein many intensity collections terminal of near infrared finger venous image comprises: led array light source, LED driving, microcontroller, near infrared camera; Said led array light source is used to provide the near infrared light of transmission finger; The LED driver module is that each LED in the led array light source provides independently constant current source driving; Micro controller module is used to control the LED driving provides drive signal to the led array light source; Control the near infrared camera simultaneously and catch the image of transmission of near infra red light finger vena radiography, micro controller module also is used for communicating with view data with PC and exchanges; The PC image processing terminal is used for the AP that the many intensity collections terminal to the near infrared finger venous image collects and merges.

Adopt the design of transmission of near infra red light finger vena contrast imaging method, according to the strong absorption characteristics of Human Physiology tissue near infrared light, red blood cell has stronger absorption characteristic for the interior near infrared light of particular range of wavelengths (690nm-980nm) in the vein; And other physiological tissues such as bone; Muscle, skin etc. to the absorption of this wave band near infrared light a little less than, the center crest that the present invention selects for use is at 850nm; Half-wave is wide just can to satisfy the wavelength requirement for the near-infrared luminous diode of 50nm; Light emitting module is rearranged by 1*N light emitting diode, and the physiological character that meets finger distributes, and the irradiation of near infrared light beam refers to the back of the body and transmission finger; Because the inboard subepidermal vein absorption of finger can the phlebography imaging.Each LED of light source module is independent controlled, and power consumption is little, and the light beam penetration power of generation is strong.

Light source module is made up of 1*8 near-infrared luminous diode.Microcontroller mainly is responsible for the control of acquisition terminal, comprise pwm control signal, near infrared camera control and with the communicating by letter of PC.Behind the absorbing model of setting up the transmission of near infra red light finger, when gathering, at first send the certain light intensity control signal of exploration; Gather finger venous image subsequently, adjust the light intensity control signal more dynamically after analyzing its intensity profile, gather finger venous image again; Continue to analyze, carry out the adjustment of light intensity control signal again, so dynamic image quality near infrared finger venous image (AP) and storage preferably of gathering some; Gathering the whole process time of these images will lack; In 0.3～0.5 second, accomplish, the user only need put into finger acquisition terminal and get final product in static about 1 second when gathering, and the image sequence that collects like this can be thought the many light intensity imagings under the Same Scene; Can think the Pixel-level registration, the Pixel-level of carrying out many light intensity image sequence again merges.Microcontroller also needs the interface that communicates with PC, and is the gate control system spare interface.

The near infrared camera selects the cmos sensor of high frame per second, and frame per second reaches 60 frames/more than second, because the near infrared finger vein image acquisition of many light intensity will be accomplished in very short time; User's finger can not move during this period, to make things convenient for registration, and need be with the fast speed images acquired; Approximately need catch image 5～8 times during many intensity collections, the time, the pixel resolution of cmos sensor reached more than the 640*480 in 0.5 second; Short, logical light quantity of the optional focal length of camera lens and the big compound glass camera lens in visual angle, the logical ranges of the band of filter plate is consistent with the near-infrared LED parameter matching, satisfies the needs that low coverage is caught finger venous image.

The present invention proposes to adopt the transmission of near infra red light finger vena radiography of multiple intensity; Use near infrared sensor (camera) to catch corresponding image; Claim that these images are the near infrared finger venous image sequence of many light intensity, therefrom select imaging effect preferably parts of images merge.

Embodiment 2

The process flow diagram of many intensity collections terminal of near infrared finger venous image when many light intensity finger vein image acquisition is as shown in Figure 2, may further comprise the steps:

(1) when beginning to gather finger venous image, microprocessor controls LED driver module sends one group of tentative light intensity control signal, for example " 20,20,20; 20,20,20,20; 20,20,20 ", expression uses the square wave of 8 20% dutycycles to drive 8 LED respectively.

(2) microcontroller obtains vein image this moment through the near infrared camera.

(3) microcontroller carries out simple image quality analysis to this vein image; Be equally divided into 10 row pieces to image by the pixel count that is listed as; Calculate the average gray value of each piece fast; If the quantity of the image column piece within suitable gray-scale value is not greater than 2 for average gray value, then looking this image is invalid image, needs to revise the light intensity control signal of corresponding row piece; If the image column number of blocks of average gray value in suitable gray-scale value is more than or equal to 8, then looking this image is AP.The present invention sets desired value and is [55,130].

For example the average gray value of certain image column piece is 30 on the low side, then when next light intensity control, sends for the LED of this piece correspondence position top " 40 " control signal; Promptly the square wave of 40% dutycycle enables to drive; And if the gray-scale value of certain image column piece is 220, then when next light intensity was controlled, the LED corresponding to this row piece sent " 10 " control signal; Promptly the square wave drive of 10% dutycycle can obtain one group of new control signal by that analogy.

(4) microcontroller judges whether to gather the AP of sufficient amount, and the present invention generally gathers 3 APs, if " denying " got into for (5) step so, if " be ", got into for (6) step so.

(5) send new light intensity control signal; Carried out for (2) step again; Obtain the finger venous image under the new light intensity condition, and then carry out (3) successively, (4) step, up to the AP that obtains sufficient amount; Perhaps experiment number reach the upper limit also directly get into (6) step, for example can be set to 30 times.

(6) collection of many light intensity finger venous image sequence is accomplished, and is stored in the pc machine AP of gained for use.

This process generally was controlled in 0.3 second, so that many light intensity of catching image sequence all is the Pixel-level registration is good.

Example: when finger is put into the finger vena acquisition terminal, begin images acquired, the test light intensity control signal that microcontroller sends for " 10,10,10,10; 10,10,10,10,10; 10 ", through driver module give 8 LED in the light source respectively with " 10%, 10%, 10%, 10%; 10%, 10%, 10%, 10% " square wave of dutycycle drives and lights, the frequency of square wave is 2MHz; The finger venous image that at this time collects such as Fig. 3 (a), this image size is 195*320 (row * row), and this image is divided into 10, shown in white piecemeal dotted line among Fig. 3 (a), can calculate the average gray (numeral under the image column piece) of each piece (195*32).It is [55,130] that the present invention in advance sets suitable image column piece average gray interval, because the average gray value of all images row piece in first image all is lower than 55, can finds out to send the first time and test that to obtain finger venous image after the light intensity be under exposed, checks the AP quantity that has photographed; Finger venous image among Fig. 3 (a) can not be regarded as AP, and 0 AP of promptly current acquisition does not satisfy 3 requirements setting, so need carry out again gathering finger venous image again after the intensity of light source adjustment; For obtaining suitable finger venous image, need adjustment light intensity grow some, microcontroller sends new light intensity control signal " 20,20; 20,20,20,20; 20,20,20,20 "; The PWM dutycycle that promptly drives 8 LED all is " 20%, 20%, 20%, 20%; 20%, 20%, 20%, 20% "; Certainly the duty cycle square wave of light intensity control can be uneven in the practical operation, can estimate the control signal of each LED next time according to the actual grey mean value of measuring for the first time, collects second finger venous image such as Fig. 3 (b), and same carries out piecemeal and ask for average gray value.After collecting the finger venous image of adjustment after the light intensity, proceed the simple analysis of image column piece average gray, have eight image column pieces to reach between suitable gray area, explain that the dutycycle of increase PWM is suitable, this image is to can be used as first effective finger venous image; Proceed the quantity of suitable images sequence below and judge, current have only 1 effective finger venous image, also needs to regulate the PWM dutycycle that drives each LED once more, for obtaining more suitably finger venous image; Need the adjustment light intensity continue grow some, microcontroller sends new light intensity control signal " 30,30,30; 30,30,30,30; 30 ", the PWM dutycycle that promptly drives 8 LED all is " 30%, 30%, 30%; 30%, 30%, 30%, 30%; 30% ", the finger venous image that obtains is seen Fig. 3 (c), repeats just now operation and can collect finger venous image again and see Fig. 3 (d), so far just having collected 3 effective finger venous images (is Fig. 3 (b); (c), (d)), be used for merging, be stored in the pc machine (perhaps microcontroller) AP of gained for use.

So far accomplished the collection of 3 effective finger venous images, these 3 images just can be gathered completion in 0.3 second, can guarantee it is that the Pixel-level registration has been got well to each other.

Embodiment 3

The Pixel-level weighting fusion method of many light intensity of near infrared finger venous image of the present invention is for comprehensive many light intensity finger venous image sequence medium sized vein lines image quality part preferably; And in fusion process, enlarge dynamic range, homogenising gray scale, contrast, and improve signal to noise ratio (S/N ratio).

The blending algorithm flow process is as shown in Figure 4:

(1) image block, the AP that above-mentioned storage is for use all carries out piecemeal, and the piecemeal rule is to be equally divided into 10 row pieces by the col width pixel count, and calculates the average gray value of each row piece.Mean value behind the image block as shown in Figure 3.

(2) image column piece quality discrimination; Promptly the image quality of each image column piece is differentiated; Because the illumination response function of optical sensor has reflected the duty relation of pixel grey scale and optical sensor, the image quality that can use this function curve to carry out the image column piece is differentiated reference literature " Recovering high dynamic range radiance maps from photographs " (P.E.Debevec and J.Malik; SIGGRAPH 97 Conf.Proc.; Computer Graphics Annual Conf.Series, pp:369-378,1997) carry out the calculating of the photosensitive response function of optical sensor.Remember that this function is B=f (I), I is the light radiation degree, and B is the grey scale pixel value of optical sensor induction, and its curve is shown in Fig. 5 (a).Can be found out by this function, be the best effort district of optical sensor at this curve middle part, that is to say that this curve derivative major part is imaging effect part preferably; Be reflected to the imaging aspect of image column piece, if promptly the average gray of image column piece drops on this curve derivative major part, this image column piece obtains when optical sensor is operated in the photosensitive interval of the best so; Just can confirm that also this image column piece is optimized image row pieces, the inverse function g of the illumination response function f of optical sensor shown in Fig. 5 (b), its derivative h=g '; Curve is shown in Fig. 5 (c), and the average gray of image column piece is designated as Bn (n=1,2; 3 ...), the longitudinal axis value of Bn correspondence on the h curve is more little; Explain that then this image column piece image quality is good more, Fig. 5 (a) and (b) in the light radiation degree has been carried out normalization, promptly minimum saturated light is 1 according to radiation intensity; The actual optimum gradation interval of choosing the image column piece is in [55,130], i.e. derivative smaller portions in Fig. 5 (c).By Fig. 5 (c) three APs (b), (c), (d) among Fig. 3 are carried out choosing of optimized image row piece, its result such as Fig. 6.

(3) weights distribute; Image quality according to the image column piece is differentiated; Can carry out the weighting weight allocation to the image column piece of various piece, the optimum branch of image quality is equipped with big weight, and all the other distribute less weight successively; Consider the block effect of eliminating on the space in addition, must be level and smooth to carrying out space transition weighting between the adjacent image row piece.Detailed step is following:

(3.1) suppose to have N to open many light intensity of near infrared finger venous image, be designated as I _n(n=1,2 ..., N), every image is divided into P image column piece, N*P image column piece so just arranged, note is I _i(i=1,2 ..., N*P), the average gray of each image column piece note is B _i, divide the informative weight value coefficient S of this image column piece of timing at weights _iCalculate by following formula (I):

S _i(x，y)＝exp[α·h(B _i)] (I)

(x, y) the row and column coordinate of remarked pixel in image, h (B in this formula _i) representing the ordinate value of average gray value Bi on Fig. 5 (c) curve of this image column piece, α is smoothing factor (getting usually between [1 0]).Promptly use the h function to decide the S of each image column piece _iWeights distribute.

(3.2) weights that only carry out the image column piece by the h function distribute not enough, are easy to generate block effect, also need carry out the level and smooth of image column interblock, and the present invention constructs one again and carries out the space smoothing coefficient G that weights distribute by distance _i, calculate by following formula (II):

$G_i(x,y)=\exp [-\frac{{(y-y_i)}^2}{2{\mathrm{\σ}}^2}]---\left(\mathrm{II}\right)$

In formula (II), σ is the standard deviation of Gaussian function, y _iBe i open under the illumination condition selected take out the central series coordinate of optimized image row piece; If a plurality of optimized image row pieces are arranged; The central series coordinate of the nearest best pixel piece of chosen distance current pixel piece then is not if all images row piece under this illumination condition is optimized image row piece, y so _iGet 0.Can confirm the space smoothing coefficient G of each pixel behind many light intensity image sequence piecemeal by this formula _i

(3.3) just can calculate final associating weights partition factor after information smoothing factor and space smoothing coefficient have been arranged, see formula (III):

w _i(x，y)＝S _i(x，y)G _i(x，y) (III)

In formula (III), w _iBe the weights partition factor of associating, can this coefficient be carried out normalization, be designated as

Promptly obtained the concrete apportioning cost of weights of each image column piece.

(4) the image column piece merges, and the occurrence that has had weights to distribute just can carry out weighting fusion with each image column piece, sees formula (IV)

$U(x,y)=\underset{i=1...N*P}{\mathrm{\Σ}}{\tilde{w}}_i(x,y)I_i(x,y)---\left(\mathrm{IV}\right)$

Image U representes the fused images of final output.

(5) syncretizing effect is differentiated, if the image averaging gray-scale value after merging between [55,130], standard deviation is less than 255, average gradient gets into step (6) greater than 10, otherwise gets into (4) step, readjusts parameter alpha and σ, carries out weighted value once more and distributes.

(6) merge end, fused images is preserved, be convenient to later image and handle application.

Example: (I) image column piece weights curve such as Fig. 7 of distributing by formula.(II) space smoothing weights curve such as Fig. 8 of distributing by formula.It is comprehensive to carry out the weights of image column piece by formula (III) then, obtains final distribution weights, like Fig. 9.Finally merge and to such an extent that image is seen Figure 10.Can be found out that by Figure 10 the gray scale of this fused images and contrast all compare evenly, merge three width of cloth image medium sized veins part preferably that forms images, the dynamic range of fused images has obtained expansion; Carried out non-linear compression after dynamic range expanded, promptly the tonal range of fused images is between [0 255], and the average gray of each row pixel of the image after the actual fused is between 55～110; The average variance of its row gray scale is far smaller than three width of cloth images before merging, and has reached the purpose of homogenising gradation of image and contrast, and the signal to noise ratio (S/N ratio) of image also is improved; If three width of cloth source images I1, I2, the power of signal and noise is respectively S1 among the I3; S2, S3 and N1, N2; N3, in the image I after merging so, the power ratio that can estimate signal S and noise N satisfies with lower inequality:

$\frac{3*\min (\mathrm{<figure-callout\; id="1"\; label="S"\; filenames="HSA00000673154200031.png,HSA00000673154200041.png"\; state="\{\{state\}\}">S</figure-callout>}1,S2,S3)}{\max (\mathrm{<figure-callout\; id="1"\; label="N"\; filenames="HSA00000673154200031.png,HSA00000673154200041.png"\; state="\{\{state\}\}">N</figure-callout>}1,N2,N3)}\&le;\frac{S}{N}\&le;\frac{3*\max (\mathrm{<figure-callout\; id="1"\; label="S"\; filenames="HSA00000673154200031.png,HSA00000673154200041.png"\; state="\{\{state\}\}">S</figure-callout>}1,S2,S3)}{\min (\mathrm{<figure-callout\; id="1"\; label="N"\; filenames="HSA00000673154200031.png,HSA00000673154200041.png"\; state="\{\{state\}\}">N</figure-callout>}1,N2,N3)}$

The signal to noise ratio (S/N ratio) Sx/Nx of any width of cloth of worthwhile ratio three width of cloth source images on this formula left side is big 2～3 times, and the method that visible the present invention proposes can effectively increase signal to noise ratio (S/N ratio).

Patent of the present invention has received the subsidy of Chongqing City's natural science fund project " Pixel-level of near infrared multispectral finger venous image merges research " (numbering cstcjjA40041).

Patent of the present invention has received the subsidy of the fund project of scientific research in the school " the many light intensity IMAQ and the fusion of near infrared finger vena " (numbering CK2011B09) of Chongqing University of Science and Technology.

Claims (3)

1. obtaining and emerging system of many light intensity of near infrared finger venous image is characterized in that, comprises the many intensity collections terminal and the PC image processing terminal of near infrared finger venous image; Wherein many intensity collections terminal of near infrared finger venous image comprises: led array light source, LED driving, microcontroller, near infrared camera; Said led array light source is used to provide the near infrared light of transmission finger; The LED driver module is that each LED in the led array light source provides independently constant current source driving; Micro controller module is used to control the LED driving provides drive signal to the led array light source; Control the near infrared camera simultaneously and catch the image of transmission of near infra red light finger vena radiography, micro controller module also is used for communicating with view data with PC and exchanges; The PC image processing terminal is used for the AP that the many intensity collections terminal to the near infrared finger venous image collects and merges.

2. obtaining and fusion method of many light intensity of near infrared finger venous image is characterized in that, comprises many light intensity finger vein image acquisition step and finger venous image fusion steps; Many light intensity finger vein image acquisition step comprises: A1, LED driver module send one group of tentative light intensity control signal; A2, microcontroller obtain vein image this moment through the near infrared camera; A3, microcontroller carry out image quality analysis to this vein image; Be equally divided into several row pieces to image by the pixel count that is listed as; Calculate the average gray value of each row piece; If the quantity of the row piece within suitable gray-scale value is not greater than 2 for average gray value, then looking this image is invalid image, needs to revise the light intensity control signal of corresponding row piece; If the row number of blocks of average gray value in suitable gray-scale value is more than or equal to 8, then looking this image is AP; A4, judge whether to gather the AP of sufficient amount, if " denying " gets into A5, if " be ", get into A6; A5, LED driver module send new light intensity control signal, and repeating step A2-A4 is up to the AP that obtains sufficient amount; A6, the collection of many light intensity finger venous image sequence are accomplished; The finger venous image fusion steps comprises: B1, image block, said AP is all carried out piecemeal, and the piecemeal rule is to be equally divided into 10 row pieces by the col width pixel count, and calculates the average gray value of each row piece; B2, image column piece quality discrimination are promptly differentiated the image quality of each image column piece; B3, weights distribute and also carry out the level and smooth of image column interblock, differentiate according to the image quality of image column piece, and the image column piece of various piece is carried out the weighting weight allocation, obtain the concrete apportioning cost of weights of each image column piece; B4, image column piece merge, and according to the concrete apportioning cost of weights, each image column piece are carried out weighting fusion; B5, syncretizing effect are differentiated; B6, fusion finish.

3. method according to claim 2 is characterized in that, said step B3 specifically carries out following steps: B31, supposes to have N to open many light intensity of near infrared finger venous image, is designated as I _n(n=1,2 ..., N), every image is divided into P image column piece, N*P image column piece so just arranged, note is I _i(i=1,2 ..., N*P), the average gray of each image column piece note is B _i, divide the informative weight value coefficient S of this image column piece of timing at weights _iCalculate by following formula (I):

S _i(x，y)＝exp[α·h(B _i)] (I)

(x, y) the row and column coordinate of remarked pixel in image, h (B in this formula _i) representing the ordinate value of average gray value Bi on Fig. 5 (c) curve of this image column piece, α is a smoothing factor;

B32, carry out the level and smooth of image column interblock, carry out the space smoothing coefficient G that weights distribute by distance _i, calculate by following formula (II):

$G_i(x,y)=\exp [-\frac{{(y-y_i)}^2}{2{\mathrm{\&sigma;}}^2}]---\left(\mathrm{II}\right)$

In formula (II), σ is the standard deviation of Gaussian function, y _iBe i open under the illumination condition selected take out the central series coordinate of optimized image row piece; If a plurality of optimized image row pieces are arranged; The central series coordinate of the nearest best pixel piece of chosen distance current pixel piece then is not if all images row piece under this illumination condition is optimized image row piece, y so _iGet 0; Can confirm the space smoothing coefficient G of each pixel behind many light intensity image sequence piecemeal by this formula _i

B33, the final associating weights partition factor of calculating, see formula (III):

w _i(x，y)＝S _i(x，y)G _i(x，y) (III)

In formula (III), w _iBe the weights partition factor of associating, can this coefficient be carried out normalization, be designated as

Promptly obtained the concrete apportioning cost of weights of each image column piece.

Images