CN105335706A - Double-frequency mixed texture fusion method - Google Patents
Double-frequency mixed texture fusion method Download PDFInfo
- Publication number
- CN105335706A CN105335706A CN201510676727.6A CN201510676727A CN105335706A CN 105335706 A CN105335706 A CN 105335706A CN 201510676727 A CN201510676727 A CN 201510676727A CN 105335706 A CN105335706 A CN 105335706A
- Authority
- CN
- China
- Prior art keywords
- palm
- image
- roi
- veins
- arteries
- 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.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/12—Fingerprints or palmprints
- G06V40/1347—Preprocessing; Feature extraction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/12—Fingerprints or palmprints
Landscapes
- Engineering & Computer Science (AREA)
- Human Computer Interaction (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Collating Specific Patterns (AREA)
- Image Processing (AREA)
Abstract
The invention relates to a double-frequency mixed texture fusion method. Based on features and the difference of changes of a texture structure in the high frequency band and the low frequency band of a frequency domain, the double-frequency mixed texture fusion method comprehensively utilizes the respective advantages in the high frequency band and the low frequency band to perfectly fuse the texture of palm print with the texture of palm venation and to achieve the effect that the fused image displaying two textures and the change rules of the two textures at the same time is displayed in the same image. The double-frequency mixed texture fusion method greatly increases the quantity of the effective features for identity recognition, and effectively improves the recognition efficiency.
Description
Technical field
The invention belongs to image identification technical field, relate generally to a kind of grain table method, particularly relate to a kind of double frequency hybrid texture fusion method.
Background technology
Palm centre of the palm texture and palm palmar metacarpal veins texture are two kinds of biological characteristics effectively identifying individual identity.Fully utilize two kinds of texture structures, both can reach higher identification accuracy, the requirement to single creature feature can also be reduced simultaneously.When fully utilizing palmmprint and palm vein reason, the biological characteristic of structure as identification of texture itself can be made full use of.Particularly, after two kinds of textures reasonably superpose, relativeness between the two will increase the quantity of the validity feature that can be used for identification greatly, improves recognition efficiency.
Conventional images fusion method is applied in two width or multiple image, there is local respectively clearly in situation, and after merging, obtaining a width, to possess the fused images of two width or the clear details of multiple image more effective simultaneously.Because palmmprint palm arteries and veins image belongs to texture image, palmmprint and the palm vein reason of same palm cross one another, and therefore existing method can not adapt to completely corresponding to the fusion of palmmprint palm arteries and veins image.
Summary of the invention
Goal of the invention
The present invention starts with for the fusion of texture type image specially, from texture structure change in the high band of frequency domain and the feature of low-frequency range and difference, comprehensive utilization high and low frequency advantage separately, palmmprint texture and palm vein reason are compared perfect fusion, reaches the fused images simultaneously presenting two kinds of textures and two kinds of texture variations rules in same sub-picture.
Technical scheme
A kind of double frequency hybrid texture fusion method, is characterized in that: the image co-registration of palmmprint palm arteries and veins for be ROI region, but not whole palm, needs the extraction and the registration that palmmprint and palm arteries and veins are carried out respectively to ROI region before merging; The method concrete steps are as follows:
(1) before carrying out ROI extraction, palm outer contour is first extracted;
(2) when obtaining palm outline, finding and referring to finger tip point, and setting up new coordinate in this, as benchmark at palmprint image, and then extracting the rectangle m obtaining palm center 256*256 in palmprint image
2region is as ROI region;
(3) arteries and veins image will be slapped using the region with palmmprint ROI same position and formed objects as slapping arteries and veins ROI.
Palm arteries and veins and palmprint image are that same palm takes acquisition continuously respectively at synchronization under 850nm near infrared light and blue light; Overall acquisition time is less than 0.1s.
No longer carry out independent ROI to palm arteries and veins image in step (3) to extract, avoid the deviation because image difference causes ROI region to extract.
Advantage and effect
This double frequency hybrid texture of the present invention fusion method, tool has the following advantages and beneficial effect:
Double frequency hybrid texture fusion method for the feature that palmmprint and palm vein are managed, considers textural characteristics and the follow-up needs carrying out identification according to texture specially.Grain distribution and trend in the texture image utilizing texture image medium-high frequency and low frequency to reflect, high-frequency information is utilized to carry out the fusion treatment of low-frequency information and high-frequency information, can reach simultaneously clear in same image and present palmmprint and palm vein reason, and the object of two kinds of texture grey scale change.Image after fusion provides abundanter effective information by for follow-up feature extraction and identification.
Accompanying drawing explanation
Fig. 1 is that palmmprint extracts image with palm arteries and veins ROI region, and wherein Fig. 1 (a) is palmmprint ROI region, and Fig. 1 (b) is palm arteries and veins ROI region.
Fig. 2 is the palm arteries and veins palmprint image after ROI synchronously extracts, and wherein Fig. 2 (a) is the palmmprint ROI image after extracting, the palm arteries and veins ROI image after Fig. 2 (b) extracts.
Fig. 3 is low frequency and vertical high frequency palmprint image after wavelet decomposition, and wherein Fig. 3 (a) is low frequency component palmprint image, and Fig. 3 (b) is vertical high fdrequency component palmprint image.
Fig. 4 is the inventive method process flow diagram.
Embodiment
The present invention is a kind of double frequency hybrid texture fusion method, palmmprint the palm arteries and veins image co-registration for be ROI region, but not whole palm, so also need the extraction and the registration that palmmprint and palm arteries and veins are carried out respectively to ROI region before merging, the method steps flow chart is as shown in Figure 4.
First, before carrying out ROI extraction, palm outer contour is first extracted.
Secondly, when obtaining palm outline, finding and referring to finger tip point, and setting up new coordinate with this method setting up new coordinate system at palmprint image, and and then extract obtain palm center 256*256 in palmprint image rectangular area as ROI region.
Finally, because palm arteries and veins and palmprint image are that same palm takes acquisition continuously respectively at synchronization under 850nm near infrared light and blue light.Overall acquisition time is less than 0.1s, and therefore in image acquisition procedures, palm does not have spatial displacement, and namely palm putting position, posture, stretching degree etc. in physical space all do not change.By palm arteries and veins image using the region with palmmprint ROI same position and formed objects as slapping arteries and veins ROI, and no longer carry out independent ROI extraction, the deviation because image difference causes ROI region to extract can be avoided like this, as shown in Figure 1, Fig. 1 (a) is palmmprint ROI region, as can be seen from the figure, the palmmprint main line of selected ROI region is more steady and audible relative to other region, and texture is also abundanter; Fig. 1 (b) is palm arteries and veins ROI region, and as can be seen from the figure under infrared imaging condition, ROI region palm arteries and veins image is clear relative to other regional structure, and texture is abundanter; Extract through ROI, and palmmprint after gray scale normalization and palm arteries and veins image are as shown in Figure 2, Fig. 2 (a) be the palmmprint ROI image after extraction, and as can be seen from the figure, palmmprint main line is relatively steady and audible, and fold line equistability is relatively poor; Palm arteries and veins ROI image after Fig. 2 (b) extracts, as can be seen from the figure under infrared imaging condition, ROI region palm arteries and veins picture structure is clear, and texture is abundanter.
If the palm arteries and veins image after ROI extraction and gray scale normalization and palmmprint input picture are designated as X1 (x, y) and X2 (x, y), x, y ∈ [1,256] respectively.
Through direction, postrotational palmprint image main line high-frequency information is mainly in vertical direction, and as shown in Figure 3, Fig. 3 (a) is low frequency component palmprint image, can see that palmmprint main line is mainly based on vertical direction from figure clearly; Fig. 3 (b) is vertical high fdrequency component palmprint image, clearly in figure sees that palmmprint main line component is more outstanding relative to non-palmmprint main line component.Therefore the information position of palmmprint main line and relative change intensity is farthest remained in the high-frequency information after vertical High frequency filter.ROI image is 256*256, and the image after three grades are decomposed is 32*32, and resolution is too low, and the quantity of information after fusion etc. have decline.Therefore only consider that firsts and seconds decomposes.Based on the vertical high-frequency information component after wavelet decomposition, complete the fusion to palmmprint main line and vein image.
First to the vertical high frequency V after palmmprint wavelet decomposition
2carry out medium filtering, namely
FV
2(x,y)=Med{V
2(x,y)}=Med[V
2(x-u,y),...,V
2(x,y),...,V
2(x+u,y)](1);
V in formula
2(x, y) for the coordinate in reconstructed image of V component after secondary wavelet decomposition be the wavelet coefficient of (x, y), x, y ∈ [1.N/4], N are the wide and high size (the wide height of handled target image is equal) of image.Choosing of u value should reach and effectively can remove noise spot, does not affect again the object of normal minutiae point, answers wz≤u<wx/2, and wz is that wx is for retention point pixel wide for removing some pixel wide.Due at V
2for removing the general wz<2 of target width in (x, y), minutiae point wx>=4, therefore getting u=1 can meet processing requirements. then by filtered vertical high frequency FV
2(x, y) component is normalized to [0,1], namely
FV
2there is positive negative in (x, y), therefore passes through FV
2(x, y)+abs (min (FV
2(x, y))) carry out numerical value translation, to eliminate negative.Palmmprint secondary wavelet decomposition low frequency coefficient X
2a
2with palm arteries and veins secondary wavelet decomposition low frequency coefficient X
1a
2coefficient fusion is carried out according to drag:
RA
2=GFV
2(x,y)×X
2A
2+(1-GFV
2(x,y)×c)×X
1A
2(3);
In formula, x, y ∈ [1, N/4], c ∈ [0,1] are coefficient of intensification, for regulating X
1a
2at the capability of influence of respective point, and be inversely proportional to capability of influence, and the larger then X of c
1a
2capability of influence is less.For guaranteeing that in the image after merging, palmmprint and vein texture information and the overall situation are abundant as much as possible, and faithful to original image, therefore choose according to the quantity of information after merging and Image Smoothness.High frequency coefficient in wavelet decomposition reflects the detailed information that palmmprint and vein comprise, and its coefficient takes the take absolute value mode of maximal value of correspondence to realize.
Wavelet inverse transformation is carried out to the wavelet coefficient after merging, the palmmprint palm arteries and veins image after merging can be obtained.
Claims (3)
1. a double frequency hybrid texture fusion method, is characterized in that: the image co-registration of palmmprint palm arteries and veins for be ROI region, but not whole palm, needs the extraction and the registration that palmmprint and palm arteries and veins are carried out respectively to ROI region before merging; The method concrete steps are as follows:
(1) before carrying out ROI extraction, palm outer contour is first extracted;
(2) when obtaining palm outline, finding and referring to finger tip point, and setting up new coordinate in this, as benchmark at palmprint image, and then extracting the rectangle obtaining palm center 256*256 in palmprint image
region is as ROI region;
(3) arteries and veins image will be slapped using the region with palmmprint ROI same position and formed objects as slapping arteries and veins ROI.
2. double frequency hybrid texture fusion method according to claim 1, is characterized in that: palm arteries and veins and palmprint image are that same palm takes acquisition continuously respectively at synchronization under 850nm near infrared light and blue light; Overall acquisition time is less than 0.1s.
3. double frequency hybrid texture fusion method according to claim 1, is characterized in that: no longer carry out independent ROI to palm arteries and veins image in step (3) and extract, avoid the deviation because image difference causes ROI region to extract.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510676727.6A CN105335706A (en) | 2015-10-16 | 2015-10-16 | Double-frequency mixed texture fusion method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510676727.6A CN105335706A (en) | 2015-10-16 | 2015-10-16 | Double-frequency mixed texture fusion method |
Publications (1)
Publication Number | Publication Date |
---|---|
CN105335706A true CN105335706A (en) | 2016-02-17 |
Family
ID=55286222
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201510676727.6A Pending CN105335706A (en) | 2015-10-16 | 2015-10-16 | Double-frequency mixed texture fusion method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN105335706A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111523413A (en) * | 2020-04-10 | 2020-08-11 | 北京百度网讯科技有限公司 | Method and device for generating face image |
Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104636721A (en) * | 2015-01-16 | 2015-05-20 | 青岛大学 | Palm print identification method based on contour and edge texture feature fusion |
-
2015
- 2015-10-16 CN CN201510676727.6A patent/CN105335706A/en active Pending
Patent Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104636721A (en) * | 2015-01-16 | 2015-05-20 | 青岛大学 | Palm print identification method based on contour and edge texture feature fusion |
Non-Patent Citations (3)
Title |
---|
李志超: "掌纹与掌静脉融合算法研究与实现", 《万方数据库》 * |
汤永华 等: "多光谱掌脉和掌纹离焦图像融合方法", 《数据采集与处理》 * |
苑玮琦 等: "基于方向梯度极值的手型轮廓跟踪算法", 《光学精密工程》 * |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111523413A (en) * | 2020-04-10 | 2020-08-11 | 北京百度网讯科技有限公司 | Method and device for generating face image |
CN111523413B (en) * | 2020-04-10 | 2023-06-23 | 北京百度网讯科技有限公司 | Method and device for generating face image |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108122213B (en) | A kind of soft image Enhancement Method based on YCrCb | |
CN106845388B (en) | Mobile terminal palm print interesting area extraction method based on complex scene | |
CN102043961B (en) | Vein feature extraction method and method for carrying out identity authentication by utilizing double finger veins and finger-shape features | |
CN104574324B (en) | The denoising method and system constrained ground groups of building remote sensing images spectral characteristic | |
CN105608409A (en) | Method and device for fingerprint identification | |
CN104091145A (en) | Human palm vein feature image acquisition method | |
CN106991380A (en) | A kind of preprocess method based on vena metacarpea image | |
CN107506688B (en) | Harris corner detection image pyramid palm print ROI identification method | |
CN104636721A (en) | Palm print identification method based on contour and edge texture feature fusion | |
CN104484652A (en) | Method for fingerprint recognition | |
CN110188639A (en) | Face image processing method and system, computer equipment and readable storage medium | |
CN108932492A (en) | A kind of image fingerprint extracting method based on non-sampled shearing wave conversion | |
Choraś | Retina recognition for biometrics | |
WO2017041552A1 (en) | Texture feature extraction method and apparatus | |
CN110569815A (en) | Finger vein image feature extraction algorithm based on improved gray level non-uniform correction | |
CN111079689B (en) | Fingerprint image enhancement method | |
Xu et al. | Direction-aware video demoireing with temporal-guided bilateral learning | |
CN105335706A (en) | Double-frequency mixed texture fusion method | |
CN109165551A (en) | A kind of expression recognition method of adaptive weighted fusion conspicuousness structure tensor and LBP feature | |
CN105528795B (en) | A kind of infrared face dividing method using annular shortest path | |
CN111611940A (en) | Rapid video face recognition method based on big data processing | |
CN106023096B (en) | Eliminate the image matching method that edge sawtoothization influences | |
Hsu et al. | Region-based image fusion with artificial neural network | |
Bastos et al. | Analysis of 2D log-Gabor filters to encode iris patterns | |
CN108550119B (en) | Image denoising method combined with edge information |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
WD01 | Invention patent application deemed withdrawn after publication | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20160217 |