CN106570895B - A kind of method for registering of optical fundus blood vessel net image - Google Patents

A kind of method for registering of optical fundus blood vessel net image Download PDF

Info

Publication number
CN106570895B
CN106570895B CN201610921209.0A CN201610921209A CN106570895B CN 106570895 B CN106570895 B CN 106570895B CN 201610921209 A CN201610921209 A CN 201610921209A CN 106570895 B CN106570895 B CN 106570895B
Authority
CN
China
Prior art keywords
block
sub
sweep
vertical
registration
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN201610921209.0A
Other languages
Chinese (zh)
Other versions
CN106570895A (en
Inventor
梁湘群
彭念
李鹏
冯东亮
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Shenzhen Moting Medical Technology Co ltd
Original Assignee
Shenzhen Certainn Technology Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Shenzhen Certainn Technology Co Ltd filed Critical Shenzhen Certainn Technology Co Ltd
Priority to CN201610921209.0A priority Critical patent/CN106570895B/en
Publication of CN106570895A publication Critical patent/CN106570895A/en
Application granted granted Critical
Publication of CN106570895B publication Critical patent/CN106570895B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00Indexing scheme for image analysis or image enhancement
    • G06T2207/30Subject of image; Context of image processing
    • G06T2207/30004Biomedical image processing
    • G06T2207/30041Eye; Retina; Ophthalmic
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00Indexing scheme for image analysis or image enhancement
    • G06T2207/30Subject of image; Context of image processing
    • G06T2207/30004Biomedical image processing
    • G06T2207/30101Blood vessel; Artery; Vein; Vascular

Abstract

The invention discloses a kind of method for registering of optical fundus blood vessel net image, it include: several laterally white bright wisps swept away described in removal in original image, by it is described sweep away original image and be divided into several first sweep away sub-block, and the removal vertical several longitudinal white bright wisps swept in original image, by it is described it is vertical sweep original image and be divided into several first vertical sweep sub-block;It is handled and successively obtains first and sweep away figure, second sweep away figure, second sweep away that figure, global registration are vertical to sweep figure;Vertical remaining the second vertical overlapping region for sweeping sub-block swept in figure of the registration process figure and the global registration is subjected to local registration one by one, obtains registration optical fundus blood vessel net image;Interpolation processing is carried out to the blank parts in the registration optical fundus blood vessel net image, obtains optical fundus blood vessel net image.This method is simple, short processing time, eliminates device optical fundus blood vessel net image collected in the distortion in transverse and longitudinal direction, has achieved the purpose that also obtain sharp picture effect when device sweeps human eye slowly.

Description

A kind of method for registering of optical fundus blood vessel net image
Technical field
The invention belongs to the technical fields that field of image processing more particularly to optical fundus blood vessel net image are registrated.
Background technique
When being scanned with ophthalmology OCT device to human eye, due to the movement that eyeball can can't help, medically it is referred to as " eye movement " collects in the figure of eyeground, it will usually the bright wisp of some whites occurs;If scanning means carries out left and right laterally to human eye It scans, then will appear several lateral white bright wisps (1A, 1B) in the blood vessel net image shown in FIG. 1 obtained;If scanning means Longitudinal scanning up and down is carried out to human eye, then will appear several longitudinal white bright wisps in the blood vessel net image shown in Fig. 2 obtained (2B).Blood vessel complete in human eye blood vessel net image is divided into several segments, is unable to accurate response eyeground by the appearance of white bright wisp The distribution situation of blood vessel makes therefore, it is necessary to which blood vessel net image shown in FIG. 1 and blood vessel net image shown in Fig. 2 to be registrated The true distribution of each blood vessel in optical fundus blood vessel net image reaction human eye after must being registrated.
Summary of the invention
The present invention provides a kind of method for registering of optical fundus blood vessel net image, and its object is to eliminate ophthalmology OCT device pair When human eye is swept slowly, the defect of the distortion generated through device optical fundus blood vessel net image collected in horizontal, longitudinal direction.
Technical scheme is as follows:
A kind of method for registering of optical fundus blood vessel net image, includes the following steps:
Human eye is scanned using ophthalmology OCT device, the original image of sweeping away for obtaining optical fundus blood vessel through processing sweeps original image with vertical;
Several laterally white bright wisps in original image are swept away described in removal, by it is described sweep away original image and be divided into several first sweep away Sub-block, and the removal vertical several longitudinal white bright wisps swept in original image, by it is described it is vertical sweep original image and be divided into several first indulge Sweep sub-block;
Maximum one piece is chosen in sub-block as vertical sweep largest block, sweep away son with every piece first from the described several first vertical sweep Block successively with it is described it is vertical sweep largest block and carry out global registration one by one, obtain several second and sweep away sub-block;All second sweep away sub-block Successively composition first sweeps away figure;
Vertical every piece second for sweeping that largest block successively sweeps away in figure with described first is swept away into sub-block and carries out local registration, It obtains several thirds and sweeps away sub-block;All thirds, which sweep away sub-block and successively form second, sweeps away figure;
It sweeps away in sub-block with the vertical largest block and described several second of sweeping and sweeps away largest block when carrying out local registration Obtained several length travels, deviate it is described it is vertical sweep largest block, obtain first and vertical sweep largest block;It is indulged with described first and sweeps largest block The described second corresponding white space swept away in figure is filled, registration process figure is obtained;
With the registration process figure be with reference to figure, by except it is described it is vertical sweep largest block in addition to remaining first vertical sweep sub-block successively Global registration is carried out with the registration process figure, several second is obtained and vertical sweeps sub-block;It is described vertical to sweep largest block and described several the Two vertical sub-blocks of sweeping successively form that global registration is vertical to sweep figure;
By vertical remaining the second vertical overlapping region one for sweeping sub-block swept in figure of the registration process figure and the global registration One carries out local registration, obtains registration optical fundus blood vessel net image;
Interpolation processing is carried out to the white space in the registration optical fundus blood vessel net image, obtains optical fundus blood vessel net image;
Wherein, the white space by it is described sweep away in original image remove several laterally white bright wisps after formed or by described It is vertical sweep remove several longitudinal white bright wisps in original image after formed.
Further: with every piece first sweep away sub-block successively with it is described it is vertical sweep largest block and carry out global registration one by one, refer to To every piece first sweep away sub-block to it is described it is vertical sweep the relevant method of largest block frequency domain, using the displacement that maximum correlation determines come Carry out whole preliminary registration.
Further: vertical every piece second for sweeping that largest block successively sweeps away in figure with described first being swept away into sub-block and is carried out Local registration obtains several thirds and sweeps away sub-block;All thirds, which sweep away sub-block and successively form second, sweeps away the specific steps of figure such as Under:
Swept away in figure from described first choose any one piece second sweep away sub-block as second to sweep away sub-block with it is described Vertical largest block of sweeping carries out local registration, obtains third and sweeps away sub-block;
By described first sweep away in figure it is remaining remaining second sweep away sub-block successively with described second to sweeping away sub-block With the vertical same method sweeping largest block and being registrated, carry out local registration with the vertical largest block of sweeping, obtain it is several its Remaining third sweeps away sub-block;
It all thirds is swept away into sub-block synthesizes described second and sweep away figure.
Further: being swept away in figure from described first and choose any one piece second and sweep away sub-block as second to sweeping away son Block and the vertical largest block of sweeping carry out local registration, obtain third and sweep away sub-block, the specific steps are as follows:
The vertical largest block and described second of sweeping is swept away sub-block and is all made of the mode of cubic spline interpolation and amplifies several times, It obtains amplifying indulging to sweep largest block and amplify second and sweeps away sub-block;
By the amplification it is vertical sweep largest block respectively crop several same number of rows pixels above and below, it is left and right respectively to crop and go The identical several columns pixel of number obtains cutting indulging and sweeps largest block;The upper and lower, left and right that sub-block is swept away in the amplification second are respectively cut out The pixel indulged with the amplification and sweep largest block same line number and columns is cut, cutting second is obtained and sweeps away sub-block;
Determine it is described cut it is vertical sweep largest block and it is described cut the second overlapping region for sweeping away sub-block, cut vertical sweep for described The overlapping region of largest block is defined as the first overlapping region, and the overlapping region that sub-block is swept away in the cutting second is defined as second Overlapping region;
First overlapping region and second overlapping region are subjected to local registration, the described secondth after registration is horizontal Sweep sub-block the second overlapping region and second sweep away sub-block Non-overlapping Domain constitute third sweep away sub-block.
Further: first overlapping region and second overlapping region being subjected to local registration, specific steps are such as Under:
It is reference with first overlapping region, by every one-row pixels of second overlapping region and first overlapping The corresponding every one-row pixels in region are registrated, and several first value of lateral displacement corresponding with number of lines of pixels are obtained;
Several first value of lateral displacement are fitted, several first lateral displacement match values are obtained;
According to several first lateral displacement match values by corresponding every one-row pixels of second overlapping region into Row translation, obtains the first registration figure;
It is reference with the first registration figure, each column pixel of first overlapping region and first registration is schemed Corresponding each column pixel is registrated, and several first longitudinal direction shift values corresponding with pixel columns are obtained;
Several first longitudinal direction shift values are fitted, several first longitudinal direction fitting shift values are obtained;
Shift values are fitted by corresponding each column pixel of first overlapping region according to several first longitudinal directions It is translated, obtains the second registration figure;
It is reference picture using the second registration figure, by every one-row pixels of the first registration figure and second registration Corresponding every one-row pixels of figure are registrated, and several second value of lateral displacement corresponding with number of lines of pixels are obtained;
Several second value of lateral displacement are fitted, several second lateral displacement match values are obtained;
According to several second lateral displacement match values by it is described first registration figure corresponding every one-row pixels into Row translation, obtains third registration figure;
Third registration figure and the second registration figure superposition are averaging, first overlapping region and described is completed Second overlapping region carries out local registration.
Further: by except it is described it is vertical sweep largest block in addition to remaining second it is vertical sweep sub-block successively with the registration process figure Global registration is carried out, refers to indulge remaining every piece described second and sweeps sub-block successively side relevant to the registration process figure frequency domain Method carries out whole preliminary registration using the displacement that maximum correlation determines.
Further: by vertical remaining the second vertical weight for sweeping sub-block swept in figure of the registration process figure and the global registration Folded region carries out local registration one by one, the optical fundus blood vessel net image being registrated, the specific steps are as follows:
Be reference picture using the registration process figure, by except it is described it is vertical sweep largest block in addition to remaining second vertical sweep sub-block one One is registrated with the registration process figure, obtains second vertical sweeping the identical several groups overlapping region of sub-block quantity with remaining;
Optional one group from the several groups overlapping region, the vertical overlapping region for sweeping sub-block of selected second will be belonged to and determined Justice is third overlapping region, and the Maximum overlap region for belonging to the registration process figure is defined as the 4th overlapping region;
By corresponding each column pixel of each column pixel of the third overlapping region and the 4th overlapping region It is registrated by column, obtains several second longitudinal direction shift values identical with overlapping region columns;
Several second longitudinal direction shift values are fitted, several second longitudinal direction displacement match values are obtained;
Match values are displaced by corresponding each column pixel of the third overlapping region according to several second longitudinal directions It is deviated, is averaging with corresponding each column pixel superposition of each column pixel and the 4th overlapping region after offset;
With pixel corresponding with the white space in each column pixel after offset, the registration process figure is filled In the corresponding white space;
Remaining remaining described second vertical sub-block of sweeping is weighed with the maximum of the registration process figure respectively with same method Folded region carries out local registration, finally obtains the registration optical fundus blood vessel net image.
Further: the Maximum overlap region of the registration process figure is the described second vertical sub-block and described of sweeping for being selected Maximum piece of overlapping area in several overlapping blocks that registration process figure is formed in registration process.
Further: the determination method of the laterally white bright wisp are as follows:
It is image I, size W*H that original image is swept away described in definition;Wherein, W indicates that image is wide, and H indicates that image is high;
It averages AI to the image data of described image I;
Longitudinal gradient transformation is carried out to described image I, obtains gradient image GI corresponding with described image I;
One mean value GI (i) is asked to every a line of the gradient image GI, wherein (1, H) i ∈, i is in described image I The laterally position of white bright wisp;
If GI (i) > a*AI, 0 is set by the position on the corresponding described image I of white bright wisp;Wherein (0,1) a ∈.
Further: the determination method of longitudinal white bright wisp are as follows:
Defining the vertical original image of sweeping is image II, size W*H;Wherein, W indicates that image is wide, and H indicates that image is high;
Transverse gradients transformation is carried out to described image II, obtains gradient image G II corresponding with described image II;
One mean value G II (j) is asked to each column of the gradient image G II, wherein j ∈ (1, W), j are described image II The position of middle longitudinal white bright wisp;
If G II (j) > a*AI, j are the position of longitudinal white bright wisp in described image II.
Beneficial effects of the present invention: the method for registering of the announced optical fundus blood vessel net image of the present invention, processing method is simple, place It is short to manage the time, in this way, eliminate ophthalmology OCT device optical fundus blood vessel net image collected in the distortion in transverse and longitudinal direction, Sharp picture effect can also be obtained when ophthalmology OCT device sweeps human eye slowly by realizing.
Detailed description of the invention
Fig. 1 sweeps away original image for optical fundus blood vessel net image;
Fig. 2 is that the vertical of optical fundus blood vessel net image sweeps original image;
Fig. 3 is the image that Fig. 1 removes that several first be divided into after several laterally white bright wisps sweep away sub-block;
Fig. 4 indicates the several first vertical images for sweeping sub-block being divided into after Fig. 2 removes longitudinal white bright wisp;
Fig. 5 indicates that selects from Fig. 4 vertical sweeps largest block 4A;
Fig. 6 indicates to remove in Fig. 3 the smallest first and sweeps away after sub-block 3D the remaining several first images for sweeping away sub-block;
Fig. 7 indicate to sweep away in sub-block and Fig. 5 by each first in Fig. 6 it is vertical sweep after largest block 4A carries out global registration, obtain Sub-block is swept away to several second, each several second, which sweep away sub-block, successively forms first and sweep away figure;
Fig. 7 .1 indicates that shown in fig. 5 indulge selected sweeps largest block 4A as registration with reference to figure;
Fig. 7 .2 indicate shown in Fig. 7 the first of selection sweep away in figure one of second sweep away sub-block 3A1;
Fig. 7 .3 indicates that vertical second swept in largest block 4A and Fig. 7 in Fig. 5 is swept away after sub-block 3A1 carries out global registration and shows Show global registration effect picture;
Sub-block 3A1 is swept away Fig. 8 shows vertical one of second swept in largest block 4A and Fig. 7 .2 in Fig. 5 to be registrated Shi Zong sweeps the overlapping region of largest block 4A;
Fig. 9 indicates that vertical second swept in largest block 4A and Fig. 7 .2 in Fig. 5 sweeps away sub-block 3A1 and carry out with punctual second cross Sweep the overlapping region of sub-block 3A1;
Figure 10 indicates vertical in the every one-row pixels and Fig. 8 of the overlapping region that in Fig. 9 second sweeps away sub-block 3A1 sweeping maximum Corresponding every one-row pixels of the overlapping region of block 4A carry out the first registration figure obtained after lateral registration;
Figure 11 indicates with Figure 10 to be with reference to figure, by each column and Figure 10 of the vertical overlapping region for sweeping largest block 4A in Fig. 8 In the corresponding each column of the first registration figure carry out the second registration figure obtained after longitudinal registration;
Figure 12 indicate with Figure 11 be refer to figure, will it is shown in Fig. 10 first be registrated figure every a line and Figure 11 shown in second Corresponding each row of registration figure carries out the third registration figure obtained after lateral registration;
Figure 13 is obtained after indicating for third registration figure shown in the second registration figure shown in Figure 11 and Figure 12 to be overlapped averagely To second sweep away sub-block registration figure;
Figure 14 indicates that vertical second swept in largest block 4A and Fig. 7 .2 in Fig. 5 sweeps away sub-block 3A1 and obtain after local registration The non-overlap part filling effect figure arrived;
Figure 15 indicates that each third sweeps away the second of sub-block registration synthesis and sweeps away figure;
Figure 16 indicate it is maximum vertical sweep block and be registrated shown in figure 15 second after offset sweep away the registration process figure that figure obtains;
Figure 17 indicates that all first vertical sweep registration process figure shown in sub-block and Figure 16 and first carry out carry out office after global registration Registration optical fundus blood vessel net image is obtained after portion's registration;
Figure 18 is indicated with the method for cubic spline interpolation to finally obtained eyeground blood after the blank parts processing in Figure 17 Pipe network image;
Figure 19 is total step of registration figure of optical fundus blood vessel net image.
Specific embodiment
In order to which technical problems, technical solutions and advantages to be solved are more clearly understood, tie below Accompanying drawings and embodiments are closed, the present invention will be described in further detail.It should be appreciated that specific embodiment described herein is only used To explain the present invention, it is not intended to limit the present invention.
With reference to Figure 19, the invention discloses a kind of method for registering of optical fundus blood vessel net image, include the following steps:
S1: scanning human eye using ophthalmology OCT device, and the original image of sweeping away for obtaining optical fundus blood vessel through processing sweeps original image with vertical;
S2: several laterally white bright wisps in original image are swept away described in removal, the original image of sweeping away is divided into several first It sweeps away sub-block, and the removal vertical several longitudinal white bright wisps swept in original image, the vertical original image of sweeping is divided into several the One vertical sweeps sub-block;
S3: maximum one piece is chosen in sub-block as vertical sweep largest block from the described several first vertical sweep, with every piece first cross Sweep sub-block successively with it is described it is vertical sweep largest block and carry out global registration one by one, obtain several second and sweep away sub-block;All second sweep away Sub-block successively forms first and sweeps away figure;
S4: it vertical every piece second for sweeping that largest block successively sweeps away in figure with described first is swept away into sub-block carries out part and match Standard obtains several thirds and sweeps away sub-block;All thirds, which sweep away sub-block and successively form second, sweeps away figure;S5: maximum is swept with described indulge Block and described several second sweeps away several length travels sweeping away largest block and obtaining when carrying out local registration in sub-block, offset It is described it is vertical sweep largest block, obtain first and vertical sweep largest block;It vertical sweep described in largest block local registration second with described first and sweeps away figure In corresponding white space, obtain registration process figure;
S6: with the registration process figure be with reference to figure, by except it is described it is vertical sweep largest block in addition to remaining first vertical sweep sub-block Global registration successively is carried out with the registration process figure, several second is obtained and vertical sweeps sub-block;If the vertical largest block and described of sweeping Dry second vertical sub-block of sweeping successively forms that global registration is vertical to sweep figure;
S7: by vertical remaining the second vertical overlapping region for sweeping sub-block swept in figure of the registration process figure and the global registration Local registration is carried out one by one, obtains registration optical fundus blood vessel net image;
S8: interpolation processing is carried out to the blank parts in the registration optical fundus blood vessel net image, obtains optical fundus blood vessel net figure Picture;
Wherein, the white space by it is described sweep away in original image remove several laterally white bright wisps after formed or by described It is vertical sweep remove several longitudinal white bright wisps in original image after formed.
For step S1 and S2, Fig. 1 to Fig. 4 is please referred to, the rete vasculosum that ophthalmology OCT device scanning human eye obtains sweeps away original Figure is shown in Fig. 1, vertical to sweep original image and see Fig. 2.In sweeping away original image, there are several laterally white bright wisps, such as 1A, 1B;Have in vertical sweep in original image Several longitudinal white bright wisps, such as 2A;Will sweep away original image and it is vertical sweep before original image is registrated, need all transverse directions in Fig. 1 White bright wisp sweeps all longitudinal white bright wisps removals in original image with vertical.When several laterally white bright wisps of removal and several longitudinal whites After bright wisp, then Fig. 1 becomes Fig. 3, and Fig. 2 just becomes Fig. 4, that is to say, that sweeps away several laterally white bright wisps of original image removal as Fig. 1 Afterwards, be divided into shown in Fig. 3 several first and sweep away sub-block, Fig. 2 it is vertical sweep several longitudinal white bright wisps of original image removal after, just divided It is indulged for shown in Fig. 4 several first and sweeps sub-block.
In order to remove several laterally white bright wisps in Fig. 1, it is necessary first to find laterally white bright wisp in Fig. 1.Look for cross To the method for white bright wisp are as follows:
S21a: enabling Fig. 1 is I, and size W*H, W indicate that image is wide, and H indicates that image is high.The image data of Fig. 1 is asked first Mean value AI.It should be noted that the size of image is W*H, referring to has the size of image section in Fig. 1, for the black in Fig. 1 Region part does not have to then consider.
S22a: longitudinal gradient transformation is carried out to image I, obtains gradient image GI (not shown) corresponding with original image;
S23a: a mean value GI (i) is asked to every a line of gradient image GI;Wherein (1, H) i ∈,
S24a: if GI (i) > a*AI, i are the position of the white bright wisp swept away in figure, by the position in bright wisp correspondence image It is set as 0, as shown in Figure 3.Wherein (0,1) a ∈.
Likewise, being also required to find in Fig. 2 first longitudinal white in order to by several longitudinal white bright wisp removals in Fig. 2 Color bright wisp.The method for looking for longitudinal white bright wisp are as follows:
S21b: enabling Fig. 2 is image II, and size W*H, W indicate that image is wide, and H indicates that image is high.It averages to image data AI.It should be noted that the size of image is W*H, referring to has the size of image section in Fig. 2, for the black region in Fig. 2 Part does not have to then consider.
S22b: transverse gradients transformation is carried out to image II, obtains gradient image G II (not shown) corresponding with image II;
S23b: a mean value G II (j) is asked to each column of gradient image G II;Wherein (1, W) j ∈;
S24b:, will longitudinal white bright wisp corresponding diagram if G II (j) > a*AI, j are the position of longitudinal white bright wisp in Fig. 2 As upper position is set as 0, as shown in Figure 4.Likewise, a ∈ (0,1).
When obtain it is shown in Fig. 3 it is several sweep away sub-block and it is shown in Fig. 4 it is several it is vertical sweep sub-block after, next just need several Sweep away sub-block and it is several it is vertical sweep sub-block and be registrated, and for the black region of the non-image portion in Fig. 3 and Fig. 4, then do not have to Participate in registration.
Step S3: choosing maximum one piece as vertical and sweep largest block from the described several first vertical sweep in sub-block, with every piece the One sweep away sub-block successively with it is described it is vertical sweep largest block and carry out global registration one by one, obtain several second and sweep away sub-block;All second It sweeps away sub-block and successively forms first and sweep away figure.
With reference to Fig. 3, after Fig. 1 is removed several laterally white bright wisps, Fig. 1 is divided into several first and sweeps away sub-block, respectively First is named as to sweep away sub-block 3A, first sweep away sub-block 3B, first sweep away sub-block 3C and first and sweep away sub-block 3D.Due to eliminating Several laterally white bright wisps, then first in Fig. 3 sweeps away sub-block 3A and first and sweeps away between sub-block 3D, first sweep away sub-block 3D and First sweeps away between sub-block 3B and first sweeps away sub-block 3B and first and sweep away between sub-block 3C and be respectively formed black horizontal stripe shape Several white spaces.Such as first sweep away sub-block 3B and first and sweep away the black horizontal stripe shape formed between sub-block 3C, as the is empty White region 3E.With reference to Fig. 6, to sweep away sub-block 3D too tiny due to first, so ignoring, therefore is removed, is formed First blank parts 3F.Then Fig. 2 is removed into longitudinal white bright wisp 2A, has obtained Fig. 4, longitudinal white bright wisp is eliminated in Fig. 4 After 2A, white space 4E is formed.Likewise, when it is shown in Fig. 2 it is vertical sweep several longitudinal white bright wisps of original image removal after, Fig. 2 just by It is divided into two piece first shown in Fig. 4 to indulge and sweeps sub-block 4A, 4B.It is vertical for especially small first to sweep sub-block (not shown), also do not join Be registrated, so being removed, leave the second blank parts (not shown).Area in Fig. 4 maximum first is indulged and sweeps sub-block 4A Be defined as it is vertical sweep largest block, 4B is then that remaining first vertical sweeps sub-block.Then, it needs to sweep away sub-block 3A for first, first sweep away Sub-block 3B, first sweep away sub-block 3C respectively and the vertical largest block 4A that sweeps carries out global registration.When every piece in Fig. 6 first sweeps away sub-block It is swept after largest block 4A successively carries out global registration with vertical, respectively obtains shown in Fig. 7 second and sweep away sub-block 3A1, second sweeps away son Block 3B1 and second sweeps away sub-block 3C1, this three piece second is swept away sub-block composition shown in Fig. 7 first and sweep away figure.It needs to illustrate Global registration described herein, refer to by first sweep away sub-block 3A, second sweep away sub-block 3B, first sweep away sub-block 3C difference To the vertical largest block 4A that sweeps using the relevant method of frequency domain, whole preliminary registration is carried out using the displacement that maximum correlation determines. Fig. 7 .1- Fig. 7 .3 specifically illustrates first and sweeps away sub-block 3A and the vertical process swept largest block 4A and carry out global registration.Fig. 7 .1 is choosing Take it is vertical sweep largest block 4A, Fig. 7 .2 is that first chosen sweeps away sub-block 3A, and Fig. 7 .3 is to sweep away sub-block 3A for first and vertical sweep most The registration process figure of bulk 4A progress global registration.
Sub-block is swept away for other first, such as: first sweeps away sub-block 3B, first sweeps away sub-block 3C, they respectively and vertical sweep The method that largest block 4A carries out global registration sweeps away sub-block 3A and the vertical method phase swept largest block 4A and carry out global registration with first Together.
Largest block 4A carry out office is swept next, sweeping away sub-block and indulging the shown in Fig. 7 first every piece second swept away in figure Portion's registration.Specifically, sub-block 3A1 is swept away by second in Fig. 7 and description is unfolded in vertical sweep for largest block 4A carries out local registration, Sub-block and the vertical method swept largest block 4A and carry out local registration are swept away for other second in Fig. 7, then sweep away sub-block with second 3A1 sweeps largest block 4A to carry out local registration identical with vertical.
Second in Fig. 7, which is illustrated, with reference to Fig. 8-Figure 14, Fig. 8 to Figure 14 sweeps away sub-block 3A1 and the vertical largest block 4A that sweeps is carried out The process of local registration.
It is inserted firstly, second in Fig. 7 .2 is swept away to indulge to sweep largest block 4A and be all made of cubic spline in sub-block 3A1 and Fig. 7 .1 The mode of value amplifies several times, obtain amplification second sweep away sub-block (not shown) and amplification indulge sweep largest block (not shown).Due to Interpolation will affect the boundary of image, so the upper and lower each several phase colleague pixels for going amplification second to sweep away sub-block are first cut out, it is left and right Each several same column pixels obtain cutting second and sweep away sub-block (not shown).Wherein, row pixel is identical with the number of column pixel. Likewise, respectively dismissing and amplify second for the vertical upper and lower, left and right for sweeping largest block of amplification sweeps away the identical line number of sub-block and columns Pixel, obtain cutting and vertical sweep largest block (not shown).
It is then determined cutting second sweeps away sub-block and cuts the vertical overlapping region for sweeping largest block, it will cut to indulge and sweep largest block Overlapping region be defined as the first overlapping region, as shown in Figure 8;The overlapping region that sub-block is swept away in cutting second is defined as second Overlapping region, as shown in Figure 9.
It should be noted that during Fig. 8 and Fig. 9 are carried out local registration, in Fig. 8 and Fig. 9 around black region Registration is not involved in calculate.The specific local registration process of Fig. 8 and Fig. 9 is as follows:
S311: being reference with the first overlapping region shown in Fig. 8, by every a line of the second overlapping region shown in Fig. 9 and The corresponding each row in first overlapping region shown in Fig. 8 is registrated, and the several first lateral position corresponding with line number is obtained Shifting value.In registration process, only considers the lateral displacement of every a line of the second overlapping region, ignore length travel.It needs to illustrate , the first value of lateral displacement is an entirety address, and every a line of the second overlapping region and the first overlapping region are corresponding Each row carry out with punctual, the first obtained value of lateral displacement should be not exactly the same.
S312: several first value of lateral displacement obtained in registration process are fitted, and obtain the several first lateral positions Move match value.It should be noted that the first lateral displacement match value is entirety address, it and the first value of lateral displacement phase It is corresponding.
S313: each row of the second overlapping region shown in Fig. 9 carried out according to several first lateral displacement match values flat It moves, obtains the first registration figure.In moving process, every one-row pixels of the second overlapping region should be with the corresponding first lateral position Shifting match value, which is used as, to be moved.After every one-row pixels when the second overlapping region complete translation, just obtain such as Figure 10 Shown in first registration figure.
S314: being reference with the first registration figure shown in Fig. 10, by each column and figure of the first overlapping region shown in Fig. 8 The corresponding each column of first registration figure shown in 10 are registrated by column, obtain several first longitudinal direction positions corresponding with columns Shifting value.In registration process, only considers the length travel of each column registration of the first overlapping region, ignore lateral displacement.It needs Illustrate, each first longitudinal direction shift value is not exactly the same.
S315: several first longitudinal direction shift values are fitted, and obtain several first longitudinal direction fitting shift values.Likewise, It is not exactly the same that each first longitudinal direction is fitted shift value.
S316: being fitted shift values according to several first longitudinal directions, by the first overlapping region shown in Fig. 8 it is each arrange into Row translation, obtains the second registration figure as shown in figure 11.It should be noted that in this step, several longitudinal fitting displacements Each column pixel of the first overlapping region is corresponding in each value and Fig. 8 of value.
S317: being reference with the second registration figure shown in Figure 11, by every a line and figure of the first registration figure shown in Fig. 10 Corresponding each row of second registration figure shown in 11 is registrated, and several second lateral displacements corresponding with line number are obtained Value;
S318: several second value of lateral displacement are fitted, and obtain several second lateral displacement match values.
S319: each traveling for being registrated image for shown in Fig. 10 first according to several second lateral displacement match values Row translation, obtains third registration figure shown in Figure 12;
S320: the second registration figure superposition shown in the registration of third shown in Figure 12 figure and Figure 11 is averaging, Figure 13 is obtained Shown in second sweep away sub-block registration figure.It should be noted that Figure 13 shows only and vertical sweep largest block 4A and second and sweep away son The overlapping region of block 3A1 carries out the effect picture obtained after local registration, sweeps largest block 4A and second for vertical and sweeps away sub-block 3A1's Non-overlap part, effect picture such as Figure 14 after overlapping region carries out local registration, at this point, lap and non-overlap part structure Sub-block Fig. 3 A2 is swept away at third.When complete it is vertical sweep largest block 4A and second and sweep away the overlapping region of sub-block 3A1 carry out part and match Standard obtains sweeping away sub-block 3B1 and 3C1 for remaining in Fig. 7 second, also after third as shown in figure 14 sweeps away sub-block Fig. 3 A2 They carry out local registration with the vertical overlapping region for sweeping largest block 4A respectively, registration thinking and vertical to sweep largest block 4A and second horizontal The thinking for sweeping the overlapping region progress local registration of sub-block 3A1 is identical, becomes corresponding third after registration respectively and sweeps away sub-block 3B2 Sub-block 3C2 is swept away with third, sees Figure 15.Finally, third sweeps away sub-block 3A2, third sweeps away sub-block 3B2 and third sweeps away sub-block 3C2 composition shown in figure 15 second sweeps away figure.
In Fig. 4 in addition to it is vertical sweep largest block 4A other than, there is also have other first it is vertical sweep sub-block, such as first vertical sweeps sub-block 4B. It should be noted that merely illustrative illustrate vertical sweeps sub-block 4A, the situation of 4B including two piece first in Fig. 4, it should be understood that First vertical to sweep the block number of sub-block be variable.
Indulge in front sweep largest block 4A with respectively sweep away the overlapping region progress local registration that sub-block be registrated during, indulge and sweep most Bulk 4A and area maximum second sweep away sub-block 3A1 (in the present patent application, the maximum block that sweeps away of area is the 3A in Fig. 7, This is only exemplary) when carrying out local registration, the vertical several length travels for sweeping largest block 4A can be obtained, it is several vertical using this Sweep largest block 4A to displacement bias is vertical, obtain first it is vertical sweep largest block 4A1, sweep largest block 4A1 filling so that first after deviating is vertical Registration process shown in figure 15 second sweeps away figure white space, obtains registration process figure shown in Figure 16.Here why emphasize Several length travels, be because offset is vertical sweep largest block 4A when, be will indulge sweep largest block 4A each column pixel carry out it is inclined It moves.A length travel can be obtained when each column pixel is deviated, since the vertical largest block 4A that sweeps includes several columns pixel, institute Several length travels can be obtained.
Each length travel mentioned here, refer to the vertical overlapping region for sweeping largest block 4A and second sweep away sub-block 3A1 ( Being that area is maximum sweeps away sub-block) the corresponding each column pixel in overlapping region carry out with obtain on time it is several after being fitted Second longitudinal direction is fitted shift value;When offset is vertical sweeps largest block 4A, refer to several second longitudinal directions fitting displacement after being fitted The vertical corresponding each column pixel for sweeping largest block 4A of value offset, and with the corresponding pixel matching of each column pixel after deviating Corresponding white space in procedure chart 15.It can be seen that each column pixel after offset only some for filling The corresponding part of white space.Herein, the mentioned-above Fig. 3 of white space leaves after removing several laterally white bright wisps Several white spaces (such as white space 3E, 3F) are formed in registration process, it may be assumed that the white space 3E in Fig. 6 is due to first Sweep away sub-block 3B and first sweep away sub-block 3C with it is vertical sweep largest block 4A experience global registration and local registration after becoming Figure 15 institute The third shown sweeps away sub-block 3B2 and after third sweeps away sub-block 3C2, becomes white space 3E1;In Fig. 6 white space 3F due to First, which sweeps away sub-block 3A and first, which sweeps away sub-block 3B, is becoming figure with vertical sweep after largest block 4A undergoes global registration and local registration Third shown in 15 sweeps away sub-block 3A2 and after third sweeps away sub-block 3B2, becomes white space 3F1.
It is reference with registration process figure shown in Figure 16, remaining in Fig. 4 first is indulged and is swept shown in sub-block 4B and Figure 16 Registration process figure elder generation global registration, obtain second it is vertical sweep sub-block (not shown), then the overlapping region that global registration obtains is carried out Local registration.It should be noted that in Fig. 4 in addition to it is vertical sweep largest block 4A other than, be left with one piece first it is vertical sweep sub-block 4B, still This is only exemplary.If there remains in Fig. 4 remaining first it is vertical sweep sub-block, need will in addition to it is vertical sweep largest block 4A other than Remaining first it is vertical sweep sub-block successively with registration process figure elder generation global registration shown in Figure 16, obtain several second it is vertical sweep sub-block, then Local registration is carried out to the several second vertical several overlapping regions for sweeping sub-block and registration process figure.
By remaining first it is vertical sweep registration process figure shown in sub-block and Figure 16 and carry out global registration when, global registration Method is also by remaining first vertical each piece for sweeping sub-block using the relevant method of frequency domain, the displacement determined using maximum correlation Whole preliminary registration successively is carried out with registration process figure shown in Figure 16 respectively.
Below by first it is vertical sweep sub-block 4B for, be specifically described first and vertical sweep how sub-block 4B was registrated with shown in Figure 16 Journey figure carries out local registration.It vertical sweep registration process figure shown in sub-block 4B and Figure 16 by first and obtains second after global registration It is vertical to sweep sub-block (not shown), in registration process and registration process figure at Maximum overlap region 5A (see Figure 17) define respectively For third overlapping region and the 4th overlapping region.Wherein, belonging to the second vertical overlapping region for sweeping sub-block is third overlapping region, is belonged to In the overlapping region of registration process figure be the 4th overlapping region.Here Maximum overlap region 5A why is selected, is because of Figure 16 Middle there is the different third of area and sweep away sub-block 3A2,3B2 and 3C2, sweeps sub-block 4B and third sweeps away sub-block 3A2,3B2 when that will indulge When carrying out local registration with 3C2,3 overlapping regions 5A, 5B and 5C shown in Figure 17 can be obtained, are needed at this time from overlapping region It is maximum as Maximum overlap region that 5A, 5B and 5C choose an area.In the present patent application, the area of overlapping region 5A is most Greatly, therefore select it as Maximum overlap region.If there are two or more areas are maximum by overlapping region 5A, 5B and 5C Overlapping region, then optional one is used as Maximum overlap region.
After selected Maximum overlap region 5A, by the corresponding of each column of third overlapping region and the 4th overlapping region Each column are registrated by column, obtain several second longitudinal direction shift values identical with overlapping region columns;By several second longitudinal directions Shift value is fitted, and obtains several second longitudinal direction displacement match values;According to several second longitudinal directions displacement match values by the Corresponding each column pixel in triple-overlapped region is deviated, to the second vertical each column pixel for sweeping sub-block be carried out inclined It moves, the second vertical sub-block of sweeping finally is become that third is vertical to sweep sub-block (not shown).In Maximum overlap region, 5A third is vertical sweeps sub-block It is superimposed and is averaging with registration process figure shown in Figure 16;Son then is swept with third is vertical for white space 3E1,3F1 shown in Figure 16 Block is filled.With same method by it is remaining remaining first vertical sweep registration process figure shown in sub-block and Figure 16 and first carry out entirely Office's registration, carries out local registration afterwards, finally obtains registration optical fundus blood vessel net image as shown in figure 17.It should be noted that this In patent application several first it is vertical to sweep sub-block be symbolistic include 4A and two pieces of 4B situation.Wherein, 4A is vertical sweeps Largest block, 4B are that smaller indulge sweeps block.Therefore, the vertical sub-block of sweeping of remaining described in this paragraph does not occur in Figure 16.In fact, First vertical sub-block of sweeping may be for 3 pieces or more, in this case in addition to first vertical sweeps that sub-block 4A and first is vertical to sweep sub-block 4B Outer also other first vertical sub-blocks of sweeping participate in first carrying out global registration with registration process figure shown in Figure 16, rear to carry out Local registration.
Finally, using the method for cubic spline interpolation, to white space (that is: the black among image of image in Figure 17 Region 5a) interpolation processing is carried out, finally obtain optical fundus blood vessel net image as shown in figure 18.White space and front are determined herein It is adopted the same.Largest block 4A and the smaller vertical situation for sweeping block 4B are swept it should be noted that only containing and indulging in Fig. 4, this is because The reason of only one longitudinal white bright wisp 2A, this is only exemplary.But it shown in Fig. 2 vertical sweep original image and is also possible to By the longitudinal white bright wisps segmentation of many items, if in this way, then Fig. 2 just will appear and very small vertical sweep sub-block.
By the above method, the registration of optical fundus blood vessel net image finally completed.The registration side of the optical fundus blood vessel net image Method, when eliminating ophthalmology OCT device and sweeping slowly to human eye, ophthalmology OCT device optical fundus blood vessel net image collected is in transverse and longitudinal direction The defect of the distortion of generation;Meanwhile the registration process of optical fundus blood vessel net image is simple, the registration time is short, realizes in ophthalmology OCT Device can also obtain the purpose of sharp picture effect in the case of sweeping slowly to human eye.
The foregoing is merely illustrative of the preferred embodiments of the present invention, is not intended to limit the invention, all in essence of the invention Made any modifications, equivalent replacements, and improvements etc., should all be included in the protection scope of the present invention within mind and principle.

Claims (9)

1. a kind of method for registering of optical fundus blood vessel net image, which comprises the steps of:
Human eye is scanned using ophthalmology OCT device, the original image of sweeping away for obtaining optical fundus blood vessel through processing sweeps original image with vertical;
Several laterally white bright wisps in original image are swept away described in removal, by it is described sweep away original image and be divided into several first sweep away son Block, and the removal vertical several longitudinal white bright wisps swept in original image, by it is described it is vertical sweep original image and be divided into several first vertical sweep Sub-block;
Choose maximum one piece in sub-block from the described several first vertical sweep as vertical and sweep largest block, with every piece first sweep away sub-block according to It is secondary with it is described it is vertical sweep largest block and carry out global registration one by one, obtain several second and sweep away sub-block;All second sweep away sub-block successively Composition first sweeps away figure;
Vertical every piece second for sweeping that largest block successively sweeps away in figure with described first is swept away into sub-block and carries out local registration, is obtained Several thirds sweep away sub-block;All thirds, which sweep away sub-block and successively form second, sweeps away figure;
Vertical largest block is swept and the described several second largest blocks of sweeping away for sweeping away in sub-block are obtained when carrying out local registration with described Several length travels, deviate it is described it is vertical sweep largest block, obtain first and vertical sweep largest block;It is filled with the described first vertical largest block of sweeping Described second sweeps away the corresponding white space in figure, obtains registration process figure;
With the registration process figure be with reference to figure, by except it is described it is vertical sweep largest block in addition to remaining first vertical sweep sub-block successively and institute It states registration process figure and carries out global registration, obtain several second and vertical sweep sub-block;The vertical largest block and described several second of sweeping is indulged It sweeps sub-block and successively forms that global registration is vertical to sweep figure;
By vertical the second vertical overlapping region for sweeping sub-block swept in figure of the registration process figure and global registration carry out office one by one Portion's registration obtains registration optical fundus blood vessel net image;
Interpolation processing is carried out to the white space in the registration optical fundus blood vessel net image, obtains optical fundus blood vessel net image;
Wherein, the white space by it is described sweep away in original image remove several laterally white bright wisps after is formed or is swept by described indulge It is formed after removing several longitudinal white bright wisps in original image.
2. the method for registering of optical fundus blood vessel net image as described in claim 1, it is characterised in that: sweep away sub-block with every piece first Successively with it is described it is vertical sweep largest block and carry out global registration one by one, refer to and sweep away sub-block to every piece first and the vertical largest block of sweeping is used The relevant method of frequency domain carries out whole preliminary registration using the displacement that maximum correlation determines.
3. the method for registering of optical fundus blood vessel net image as described in claim 1, it is characterised in that: by it is described it is vertical sweep largest block according to It is secondary to sweep away sub-block progress local registration with the described first every piece second swept away in figure, it obtains several thirds and sweeps away sub-block;It is all Third, which sweeps away sub-block and successively forms second, sweeps away figure specific step is as follows:
Swept away in figure from described first choose any one piece second sweep away sub-block as second to sweep away sub-block and it is described indulge sweep Largest block carries out local registration, obtains third and sweeps away sub-block;
By described first sweep away in figure it is remaining remaining second sweep away sub-block successively with described second to sweeping away sub-block and institute State it is vertical sweep the same method that largest block is registrated, carry out local registration with the vertical largest block of sweeping, obtain it is several remaining the Three sweep away sub-block;
It all thirds is swept away into sub-block synthesizes described second and sweep away figure.
4. the method for registering of optical fundus blood vessel net image as claimed in claim 3, it is characterised in that: swept away in figure from described first It chooses any one piece second and sweeps away sub-block as second to obtain with sub-block is swept away and the vertical largest block of sweeping carries out local registration Third sweeps away sub-block, the specific steps are as follows:
The vertical largest block and described second of sweeping is swept away sub-block and is all made of the mode of cubic spline interpolation and amplifies several times, is obtained Amplification, which is indulged to sweep largest block and amplify second, sweeps away sub-block;
By the amplification it is vertical sweep largest block respectively crop several same number of rows pixels above and below, it is left and right respectively crop it is several identical Columns pixel obtains cutting indulging and sweeps largest block, and the line number reduced is identical as the columns reduced;The amplification second is swept away Sub-block respectively crops several same number of rows pixels above and below, left and right respectively to crop several same number of columns pixels, obtains cutting the Two sweep away sub-block, and the amplification is vertical to sweep line number that largest block is reduced and described amplify second and sweep away the line number phase that sub-block is reduced Together, the amplification is vertical sweeps columns that largest block is reduced and amplifies second with described to sweep away the columns that sub-block is reduced identical;
Determine it is described cut it is vertical sweep largest block and it is described cut the second overlapping region for sweeping away sub-block, described cut vertical is swept into maximum The overlapping region of block is defined as the first overlapping region, and the overlapping region that sub-block is swept away in the cutting second is defined as the second overlapping Region;
First overlapping region and second overlapping region are subjected to local registration, described second after registration sweeps away sub-block The second overlapping region and second sweep away sub-block Non-overlapping Domain constitute third sweep away sub-block.
5. the method for registering of optical fundus blood vessel net image as claimed in claim 4, it is characterised in that: by first overlapping region Local registration is carried out with second overlapping region, the specific steps are as follows:
It is reference with first overlapping region, by every one-row pixels of second overlapping region and first overlapping region Corresponding every one-row pixels are registrated, and several first value of lateral displacement corresponding with number of lines of pixels are obtained;
Several first value of lateral displacement are fitted, several first lateral displacement match values are obtained;
Corresponding every one-row pixels of second overlapping region are carried out according to several first lateral displacement match values flat It moves, obtains the first registration figure;
It is reference with the first registration figure, each column pixel of first overlapping region and the first registration figure is opposite The each column pixel answered is registrated, and several first longitudinal direction shift values corresponding with pixel columns are obtained;
Several first longitudinal direction shift values are fitted, several first longitudinal direction fitting shift values are obtained;
Corresponding each column pixel of first overlapping region is carried out according to several first longitudinal direction fitting shift values Translation, obtains the second registration figure;
It is reference picture using the second registration figure, every one-row pixels and described second of the first registration figure is registrated figure Corresponding every one-row pixels are registrated, and several second value of lateral displacement corresponding with number of lines of pixels are obtained;
Several second value of lateral displacement are fitted, several second lateral displacement match values are obtained;
Corresponding every one-row pixels of the first registration figure are carried out according to several second lateral displacement match values flat It moves, obtains third registration figure;
Third registration figure and the second registration figure superposition are averaging, first overlapping region and described second are completed Overlapping region carries out local registration.
6. the method for registering of optical fundus blood vessel net image as described in claim 1, it is characterised in that: largest block will be swept except described indulge Remaining first vertical sub-block of sweeping in addition successively carries out global registration with the registration process figure, refers to remaining every piece described first Vertical to sweep sub-block successively method relevant to the registration process figure frequency domain, the displacement determined using maximum correlation is whole to carry out Body preliminary registration.
7. the method for registering of optical fundus blood vessel net image as described in claim 1, it is characterised in that: by the registration process figure and Vertical the second vertical overlapping region for sweeping sub-block swept in figure of the global registration carries out local registration one by one, the eyeground blood being registrated Pipe network image, the specific steps are as follows:
Be reference picture using the registration process figure, by except it is described it is vertical sweep largest block in addition to remaining second it is vertical sweep sub-block one by one with The registration process figure is registrated, and obtains second vertical sweeping the identical several groups overlapping region of sub-block quantity with remaining;
Optional one group from the several groups overlapping region, the vertical overlapping region for sweeping sub-block of selected second will be belonged to and be defined as Third overlapping region, the Maximum overlap region for belonging to the registration process figure are defined as the 4th overlapping region;
Corresponding each column pixel of each column pixel of the third overlapping region and the 4th overlapping region is carried out It is registrated by column, obtains several second longitudinal direction shift values identical with overlapping region columns;
Several second longitudinal direction shift values are fitted, several second longitudinal direction displacement match values are obtained;
Corresponding each column pixel of the third overlapping region is carried out according to several second longitudinal direction displacement match values Offset is averaging with corresponding each column pixel superposition of each column pixel and the 4th overlapping region after offset;
With pixel corresponding with the white space in each column pixel after offset, fill in the registration process figure The corresponding white space;
With same method by it is remaining it is described remaining second vertical sweep Maximum overlap area of the sub-block respectively with the registration process figure Domain carries out local registration, finally obtains the registration optical fundus blood vessel net image.
8. the method for registering of optical fundus blood vessel net image as claimed in claim 7, it is characterised in that: the registration process figure is most If big overlapping region is that described second selected indulges and sweep sub-block and dry weight that the registration process figure is formed in registration process Maximum piece of overlapping area in folded block.
9. such as the method for registering of optical fundus blood vessel net image of any of claims 1-8, it is characterised in that: the cross To the determination method of white bright wisp are as follows:
It is image I, size W*H that original image is swept away described in definition;Wherein, W indicates that image is wide, and H indicates that image is high;
It averages AI to the image data of described image I;
Longitudinal gradient transformation is carried out to described image I, obtains gradient image GI corresponding with described image I;
One mean value GI (i) is asked to every a line of the gradient image GI, wherein (1, H) i ∈, i is the transverse direction in described image I The position of white bright wisp;
If GI (i) > a*AI, 0 is set by the position on the corresponding described image I of white bright wisp;Wherein (0,1) a ∈.
CN201610921209.0A 2016-10-21 2016-10-21 A kind of method for registering of optical fundus blood vessel net image Active CN106570895B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201610921209.0A CN106570895B (en) 2016-10-21 2016-10-21 A kind of method for registering of optical fundus blood vessel net image

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201610921209.0A CN106570895B (en) 2016-10-21 2016-10-21 A kind of method for registering of optical fundus blood vessel net image

Publications (2)

Publication Number Publication Date
CN106570895A CN106570895A (en) 2017-04-19
CN106570895B true CN106570895B (en) 2019-04-26

Family

ID=58534033

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201610921209.0A Active CN106570895B (en) 2016-10-21 2016-10-21 A kind of method for registering of optical fundus blood vessel net image

Country Status (1)

Country Link
CN (1) CN106570895B (en)

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102458222A (en) * 2009-06-02 2012-05-16 佳能株式会社 Image processing apparatus, image processing method, and computer program
CN104411233A (en) * 2012-02-03 2015-03-11 俄勒冈健康科学大学 In vivo optical flow imaging
CN104768446A (en) * 2012-09-10 2015-07-08 俄勒冈健康科学大学 Quantification of local circulation with OCT angiography
CN105686795A (en) * 2016-01-13 2016-06-22 深圳市斯尔顿科技有限公司 En face OCT image dynamic display method
CN105796054A (en) * 2016-02-19 2016-07-27 深圳市斯尔顿科技有限公司 OCT image processing method and device

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102458222A (en) * 2009-06-02 2012-05-16 佳能株式会社 Image processing apparatus, image processing method, and computer program
CN104411233A (en) * 2012-02-03 2015-03-11 俄勒冈健康科学大学 In vivo optical flow imaging
CN104768446A (en) * 2012-09-10 2015-07-08 俄勒冈健康科学大学 Quantification of local circulation with OCT angiography
CN105686795A (en) * 2016-01-13 2016-06-22 深圳市斯尔顿科技有限公司 En face OCT image dynamic display method
CN105796054A (en) * 2016-02-19 2016-07-27 深圳市斯尔顿科技有限公司 OCT image processing method and device

Also Published As

Publication number Publication date
CN106570895A (en) 2017-04-19

Similar Documents

Publication Publication Date Title
CN101065772B (en) Multi-feature time filtering for enhancing structures in noisy images
CN103764036B (en) Medical diagnosis image processing apparatus
EP1074001B1 (en) Image processing method, system and apparatus for forming an overview image of an elongated scene
JP2007160094A (en) Method and apparatus for visualizing series of image data set by tomography
CN103886621B (en) A kind of method for automatically extracting bed board
CN105246405A (en) Medical image processing device and X-ray diagnostic device
CN100490747C (en) 3-dimensional image processing apparatus
CN107106102B (en) Digital subtraction angiography
JPH1057365A (en) X-ray imaging method
ES2706749T3 (en) Method to process image data that represent a three-dimensional volume
CN104517303A (en) Medical image processing apparatus and medical image processing method
JP5600285B2 (en) Ultrasonic image processing device
CN105913479B (en) A kind of vessel surface method for reconstructing based on cardiac CT image
CN107656224A (en) A kind of MR imaging method, apparatus and system
CN110111258A (en) Infrared excess resolution reconstruction image method and system based on multi-core processor
US6366635B1 (en) Method and apparatus for three-dimensional image-rendering of a spatial and tissue-based configuration through separating high contrast and injected contrast agents in multi-angular x-ray absorption measurement
CN106570895B (en) A kind of method for registering of optical fundus blood vessel net image
CN105009581A (en) Stereoscopic endoscope system
CN110084805B (en) FOV parameter setting method and device and image processing equipment
CN101213573A (en) Image processing
KR20110136487A (en) Apparatus and method for obtaining medical fusion image
CN105631908A (en) PET image reconstruction method and device
CN113936069B (en) Array element virtual interpolation method for photoacoustic tomography
JP6640550B2 (en) X-ray diagnostic equipment
CN206649468U (en) Adaptive dynamic solid augmented reality operation navigation system based on real-time tracking and Multi-source Information Fusion

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant
CP03 Change of name, title or address

Address after: 518000 803, block B, Jingang center, Jingang building, houye community, Xixiang street, Bao'an District, Shenzhen, Guangdong

Patentee after: Shenzhen moting Medical Technology Co.,Ltd.

Address before: 518112 Room 501, block C, building 2, saitu Science Park, No. 137, Bulan Road, Buji street, Longgang District, Shenzhen, Guangdong Province

Patentee before: SHENZHEN CERTAINN TECHNOLOGY Co.,Ltd.

CP03 Change of name, title or address
CP03 Change of name, title or address

Address after: Room L302, Building 2, Skyworth Innovation Valley, No. 8 Tangtou 1st Road, Tangtou Community, Shiyan Street, Bao'an District, Shenzhen City, Guangdong Province, 518108

Patentee after: Shenzhen Moting Medical Technology Co.,Ltd.

Address before: 518000 803, block B, Jingang center, Jingang building, houye community, Xixiang street, Bao'an District, Shenzhen, Guangdong

Patentee before: Shenzhen moting Medical Technology Co.,Ltd.

CP03 Change of name, title or address