CN102737230B - Non-local mean filtering method based on direction field estimation - Google Patents
Non-local mean filtering method based on direction field estimation Download PDFInfo
- Publication number
- CN102737230B CN102737230B CN201210166669.9A CN201210166669A CN102737230B CN 102737230 B CN102737230 B CN 102737230B CN 201210166669 A CN201210166669 A CN 201210166669A CN 102737230 B CN102737230 B CN 102737230B
- Authority
- CN
- China
- Prior art keywords
- fingerprint image
- pixel
- sigma
- local mean
- block
- 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
Links
Images
Landscapes
- Collating Specific Patterns (AREA)
Abstract
The invention discloses a non-local mean filtering method based on direction field estimation. The non-local mean filtering method comprises the following steps of receiving a discrete noise fingerprint image, establishing a direction field estimation model of a pixel block in the discrete noise fingerprint image, and performing non-local mean filtering on the discrete noise fingerprint image based on the direction field estimation model to obtain a final denoising fingerprint image. According to the non-local mean filtering method, the problem of poor inhibition performance and poor robustness on noise in the conventional method is solved, contrast among grains in the fingerprint image is enhanced, and characteristic information in the fingerprint image is protected from damage.
Description
Technical field
The invention belongs to fingerprint image denoising and strengthen field, more specifically, relate to a kind of non-local mean filtering method based on orientation estimate.
Background technology
Along with the develop rapidly of biological identification technology in recent years, fingerprint recognition is stable with it, and efficient, advantage becomes the identity recognizing technology of a widespread use easily.In fingerprint recognition system, algorithm for recognizing fingerprint is one of focus of wherein studying, particularly aspect the denoising enhancing of noise fingerprint image.It is indispensable operation in fingerprint image preprocessing that denoising strengthens, and its treatment effect directly affects the validity and reliability of subsequent fingerprint the matching analysis.But owing to being subject to the impact of the factors such as finger surface cleanliness, environment-identification complexity and the defect of fingerprint image sensor own, the fingerprint image obtaining has all been subject to serious noise corrosion conventionally, has directly affected the information extraction of fingerprint image.
Non-local mean filtering is mainly proposed by people such as Buades, this algorithm utilization be the global information of image, adopt Gauss's weight Euclidean distance of block of pixels to weigh the similarity between pixel, with this, come in Recovery image by the pixel of noise pollution, and traditional part filter method has only been utilized near the local message pending point on image, compare traditional part filter method, non-local mean filtering can be removed the noise in image well.
But there is following defect in the denoising enhancing field of noise fingerprint image in traditional non-local mean filtering:
1) not strong to the rejection of noise, poor robustness;
2) reduced the contrast between lines in fingerprint image;
3) corroded the characteristic information in fingerprint image.
Summary of the invention
For the defect of prior art; the object of the present invention is to provide a kind of non-local mean filtering method based on orientation estimate; be intended to solve the problem of exist in existing method strong to the rejection of noise, poor robustness; improve the contrast between lines in fingerprint image, and protect the characteristic information in fingerprint image not to be corroded.。
For achieving the above object, the invention provides a kind of non-local mean filtering method based on orientation estimate, comprise the following steps:
(1) receive a width discrete noise fingerprint image I, and set up the orientation estimate model of block of pixels in discrete noise fingerprint image;
If pixel i arbitrarily, j ∈ I, the field of direction of the block of pixels centered by pixel i is estimated by following equation:
Wherein (u, is v) the coordinate of any pixel, and (m, n) is the coordinate of pixel i, G
x(u, v) and G
y(u, v) is respectively any pixel coordinate (u, the horizontal gradient of v) locating and VG (vertical gradient), V
x(m, n) and V
y(m, n) be respectively centered by (m, n), size estimates and VG (vertical gradient) estimation for the horizontal gradient of the block of pixels of s, θ
ifor the field of direction of the block of pixels centered by pixel i;
(2) based on orientation estimate model, discrete noise fingerprint image is carried out to non-local mean filtering, to obtain final denoising fingerprint image, specifically comprises following sub-step:
(2-1) according to following equation, calculate normalized factor:
Wherein, C(i) be normalized factor, N
kthe rectangular block of pixels centered by pixel k, p (N
k) be block of pixels N
kgray level vector, h
1and h
2for decay factor, its value is determined by experiment, for the attenuation degree of control characteristic function;
(2-2) according to normalized factor C(i) calculate similarity weight, specifically adopt following formula:
Wherein w(i, j) be similarity weight;
(2-3) according to similarity weight, calculate final denoising fingerprint image, particularly, adopt following formula:
Wherein ANL[I] (i) be final denoising fingerprint image.
Horizontal gradient and VG (vertical gradient) are to calculate by Sobel operator.
The above technical scheme of conceiving by the present invention, compared with prior art, the present invention has following beneficial effect:
1, the field of direction that the present invention adopts has improved the similarity evaluation level between block of pixels, avoided the similar but dissimilar block of pixels of structure of gray scale to be incorporated in filtering, thereby cause the increase of method noise, therefore the present invention has stronger robustness to noise fingerprint image, can obviously suppress picture noise, significantly improve the PSNR value of image;
2, in filtering of the present invention, adopt the exponential function impact of the low block of pixels of similarity on point-of-interest that decay, because the rate of decay of exponential function is very fast, nuance between block of pixels can cause larger decay, thereby the gray difference between different lines is increased, therefore make the Enhancement contrast between lines in fingerprint image;
3, the structural information in the direction field energy recognition image in the present invention, and similarity weight in filter operator comprises field of direction difference, therefore, in filtering, the structural information of unique point is fully retained by the impact of weight.
Accompanying drawing explanation
Fig. 1 is the process flow diagram that the present invention is based on the non-local mean filtering method of orientation estimate.
Fig. 2 (a) to (d) is the test result comparison of 256 × 256 fingerprint simulation figure.
Fig. 3 (a) to (d) is the test result comparison of actual fingerprint collection image.
Embodiment
In order to make object of the present invention, technical scheme and advantage clearer, below in conjunction with drawings and Examples, the present invention is further elaborated.Should be appreciated that specific embodiment described herein, only in order to explain the present invention, is not intended to limit the present invention.
As shown in Figure 1, the non-local mean filtering method that the present invention is based on orientation estimate comprises the following steps:
(1) receive a width discrete noise fingerprint image I, and set up the orientation estimate model of block of pixels in discrete noise fingerprint image;
To this discrete noise fingerprint image I, establish pixel i arbitrarily, j ∈ I, (u, is v) the coordinate of any pixel, and the field of direction of the block of pixels centered by pixel i is estimated by following equation:
Wherein (m, n) is the coordinate of pixel i, G
x(u, v) and G
y(u, v) is respectively that (u, the horizontal gradient of v) locating and VG (vertical gradient) are calculated V to pixel by sobel operator
x(m, n) and V
y(m, n) is respectively that size is horizontal gradient estimation and the VG (vertical gradient) estimation of the block of pixels of s, θ centered by (m, n)
ifor the field of direction of the block of pixels centered by pixel i.
(2) based on orientation estimate model, discrete noise fingerprint image is carried out to non-local mean filtering, to obtain final denoising fingerprint image;
Particularly, this step comprises following sub-step:
(2-1) according to following equation, calculate normalized factor:
Wherein, C(i) be normalized factor, N
kthe rectangular block of pixels centered by pixel k, p (N
k) be block of pixels N
kgray level vector, h
1and h
2for decay factor, its value is determined by experiment, for the attenuation degree of control characteristic function;
(2-2) according to normalized factor C(i) calculating similarity weight;
The following formula of concrete employing:
Wherein w(i, j) be similarity weight;
(2-3) according to similarity weight, calculate final denoising fingerprint image;
Particularly, adopt following formula:
Wherein ANL[I] (i) be final denoising fingerprint image.
In the present invention, we are incorporated into field of direction similarity in the estimation of similarity weight, similarity weight is after revising, not only can reflect the gray level similarity of pixel interblock, and can also accurately reflect by directional information the texture structure characteristic of fingerprint image, thereby can better carry out fingerprint image denoising.
As shown in Figure 2, the fingerprint simulation figure that employing size is 256 × 256 is as test pattern.In this example, h
1=240, h
2=30, s=5.Fig. 2 (a) is untainted fingerprint simulation figure, Fig. 2 (b) noise pattern after for 40 the Gaussian noise of having added that standard deviation is, Fig. 2 (c) is with the recovery figure after the denoising of traditional non-local mean filtering algorithm, and Fig. 2 (d) is for adopting the recovery figure after the non-local mean filtering method denoising that the present invention is based on orientation estimate.In addition, we have added respectively standard deviation sigma to test pattern is in this example 40,50, after 60 Gaussian noise, respectively the fingerprint simulation figure that has added noise is carried out to denoising with traditional non-local mean filtering algorithm and the non-local mean filtering algorithm based on orientation estimate, and adopt herein Y-PSNR (PSNR) to weigh the non-local mean filtering algorithm based on orientation estimate that we propose and the denoising effect of traditional non-local mean filtering algorithm.Following table has shown comparative result:
Standard deviation sigma is larger, and noise is stronger to the extent of corrosion of image, and PSNR value shows that more greatly the image after denoising approaches original emulating image, and denoising effect is better.Visible by upper table, the non-local mean filtering method based on orientation estimate of the present invention is all better than traditional non-local mean filtering to the denoising ability of image in the situation that of different noise corrosion.
As shown in Figure 3, actual fingerprint collecting image is used as tested object, wherein h
1=250, h
2=20, s=5.Fig. 3 (a) is actual fingerprint collecting image, Fig. 3 (b) is the enlarged drawing of image in white marking frame in Fig. 3 (a), Fig. 3 (c) is with the recovery figure after the denoising of traditional non-local mean filtering algorithm, and Fig. 3 (d) is with the recovery figure after the non-local mean filtering method denoising based on orientation estimate proposed by the invention.By naked eyes, can obviously observe the non-local mean filtering algorithm based on orientation estimate proposed by the invention and be better than traditional non-local mean filtering algorithm aspect removal fingerprint image noise and reservation fingerprint image structural information.
Those skilled in the art will readily understand; the foregoing is only preferred embodiment of the present invention; not in order to limit the present invention, all any modifications of doing within the spirit and principles in the present invention, be equal to and replace and improvement etc., within all should being included in protection scope of the present invention.
Claims (2)
1. the non-local mean filtering method based on orientation estimate, is characterized in that, comprises the following steps:
(1) receive a width discrete noise fingerprint image I, and set up the orientation estimate model of block of pixels in discrete noise fingerprint image;
If pixel i arbitrarily, j ∈ I, the field of direction of the block of pixels centered by pixel i is estimated by following equation:
Wherein (u, is v) the coordinate of any pixel, and (m, n) is the coordinate of pixel i, G
x(u, v) and G
y(u, v) is respectively any pixel coordinate (u, the horizontal gradient of v) locating and VG (vertical gradient), V
x(m, n) and V
y(m, n) be respectively centered by (m, n), size estimates and VG (vertical gradient) estimation for the horizontal gradient of the block of pixels of s, θ
ifor the field of direction of the block of pixels centered by pixel i;
(2) based on orientation estimate model, discrete noise fingerprint image is carried out to non-local mean filtering, to obtain final denoising fingerprint image, specifically comprises following sub-step:
(2-1) according to following equation, calculate normalized factor:
Wherein, C(i) be normalized factor, N
kthe rectangular block of pixels centered by pixel k, p (N
k) be block of pixels N
kgray level vector, h
1and h
2for decay factor, its value is determined by experiment, for the attenuation degree of control characteristic function;
(2-2) according to normalized factor C(i) calculate similarity weight, specifically adopt following formula:
Wherein w(i, j) be similarity weight;
(2-3) according to similarity weight, calculate final denoising fingerprint image, particularly, adopt following formula:
Wherein ANL[I] (i) be final denoising fingerprint image.
2. method according to claim 1, is characterized in that, horizontal gradient and VG (vertical gradient) are to calculate by Sobel operator.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201210166669.9A CN102737230B (en) | 2012-05-25 | 2012-05-25 | Non-local mean filtering method based on direction field estimation |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201210166669.9A CN102737230B (en) | 2012-05-25 | 2012-05-25 | Non-local mean filtering method based on direction field estimation |
Publications (2)
Publication Number | Publication Date |
---|---|
CN102737230A CN102737230A (en) | 2012-10-17 |
CN102737230B true CN102737230B (en) | 2014-05-07 |
Family
ID=46992699
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201210166669.9A Active CN102737230B (en) | 2012-05-25 | 2012-05-25 | Non-local mean filtering method based on direction field estimation |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN102737230B (en) |
Families Citing this family (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104077560B (en) * | 2014-01-13 | 2017-07-04 | 北京市公安局刑事侦查总队 | Fingerprint comparison method |
CN105125204B (en) * | 2015-07-31 | 2017-11-17 | 华中科技大学 | A kind of electrocardiosignal noise-reduction method based on ESMD methods |
CN106407920B (en) * | 2016-09-07 | 2019-08-23 | 深圳芯启航科技有限公司 | The fringes noise removing method of fingerprint image |
CN107389516B (en) * | 2017-07-17 | 2019-10-29 | 江苏锡沂高新区科技发展有限公司 | A kind of efficient factory floor dust monitoring system |
CN107341495A (en) * | 2017-07-17 | 2017-11-10 | 深圳森阳环保材料科技有限公司 | A kind of effective underground coal dust detecting system |
CN111046893B (en) * | 2018-10-12 | 2024-02-02 | 富士通株式会社 | Image similarity determining method and device, image processing method and device |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101833753B (en) * | 2010-04-30 | 2012-05-09 | 西安电子科技大学 | SAR image de-speckling method based on improved Bayes non-local mean filter |
-
2012
- 2012-05-25 CN CN201210166669.9A patent/CN102737230B/en active Active
Also Published As
Publication number | Publication date |
---|---|
CN102737230A (en) | 2012-10-17 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN102737230B (en) | Non-local mean filtering method based on direction field estimation | |
CN107808378B (en) | Method for detecting potential defects of complex-structure casting based on vertical longitudinal and transverse line profile features | |
CN102880877B (en) | Target identification method based on contour features | |
CN101551863B (en) | Method for extracting roads from remote sensing image based on non-sub-sampled contourlet transform | |
CN103679118A (en) | Human face in-vivo detection method and system | |
CN103218605B (en) | A kind of fast human-eye positioning method based on integral projection and rim detection | |
CN105405142A (en) | Edge defect detection method and system for glass panel | |
CN104424457A (en) | Method for identifying two-dimensional code under the condition of nonlinear distortion | |
CN102682428B (en) | Fingerprint image computer automatic mending method based on direction fields | |
CN104346619A (en) | Device to extract biometric feature vector, method to extract biometric feature vector and program to extract biometric feature vector | |
CN109919002B (en) | Yellow stop line identification method and device, computer equipment and storage medium | |
US20180174328A1 (en) | Turning radius-based corner detection algorithm | |
CN102819733A (en) | Rapid detection fuzzy method of face in street view image | |
CN107066961A (en) | Fingerprint method for registering and device | |
CN108229247A (en) | A kind of mobile vehicle detection method | |
CN102073872A (en) | Image-based method for identifying shape of parasite egg | |
CN103914829A (en) | Method for detecting edge of noisy image | |
CN109948489A (en) | A kind of face identification system and method based on the fusion of video multiframe face characteristic | |
CN104680148B (en) | A kind of Fingerprint diretion computational methods based on improved Partial Differential Equation Model | |
CN104063682A (en) | Pedestrian detection method based on edge grading and CENTRIST characteristic | |
CN111950606B (en) | Knife switch state identification method, device, equipment and storage medium | |
CN109389644B (en) | Direction gradient enhancement-based parking space line detection method | |
CN105205829A (en) | Transformer substation infrared image segmentation method based on improved two-dimensional Otsu algorithm | |
CN109101874A (en) | A kind of library's robot barrier recognition methods based on depth image | |
CN108009480A (en) | A kind of image human body behavioral value method of feature based identification |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C14 | Grant of patent or utility model | ||
GR01 | Patent grant |