CN106611424A - Image edge extraction method - Google Patents
Image edge extraction method Download PDFInfo
- Publication number
- CN106611424A CN106611424A CN201610173046.2A CN201610173046A CN106611424A CN 106611424 A CN106611424 A CN 106611424A CN 201610173046 A CN201610173046 A CN 201610173046A CN 106611424 A CN106611424 A CN 106611424A
- Authority
- CN
- China
- Prior art keywords
- image
- gradient
- edge
- high frequency
- frequency subgraph
- 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
Landscapes
- Image Processing (AREA)
- Image Analysis (AREA)
Abstract
The invention provides an image edge extraction method. According to the method, a two-dimensional nonseparable lowpass filter is constructed; first an original image is subjected to multi-scale nonseparable additive wavelet decomposition; then a gradient map of decomposed sub-images is obtained by utilizing a modern mathematical morphology gradient operator; high-frequency sub-images are subjected to addition, so that the edges of low-frequency and high-frequency sub-images are extracted; and a final edge image is formed again. A new image edge extraction method is provided by utilizing characteristics of richness, quickness and the like of two-dimensional nonseparable additive wavelet isotropy and morphology gradient detection of the edges. The method is simple in operation and good in detection effect; a target with a relatively small gray value in the image can be detected, more edge details can be obtained, and the edges with the integrity, multi-directionality, continuity and translation invariance can be extracted. Compared with other detection methods, the method has a relatively high processing speed.
Description
Art
The present invention relates to the field such as target recognition, image segmentation, remote sensing, medical image analysis.
Background technology
Marginal information is extremely important information in image, in theory, the whole in image can be recovered by marginal information
Information.Thus rim detection is the important content of graphical analyses；It is the key for processing many problems；Traditional rim detection is main
It is to detect edge horizontally and vertically respectively with difference operator horizontally and vertically, then synthesizes
Certain gradient carries out rim detection, and a demand goes out the difference of both direction when computer is realized, then synthesizes, but this side
Method also has clearly disadvantageous, and they only emphasize to be included in level, vertical, the marginal information of both direction, but general real image
There is the marginal information of multiple directions even any direction.
Mathematical morphology is the powerful of analysis of the image geometric properties, it by with some basic set operations,
Such as burn into expansion, open and close are analyzed process to picture shape and structure, in the image procossing of such as image enhaucament
Good application is arrived, the Morphological Gradient of burn into dilation operation has preferable Image Edge-Detection effect in morphology, but
Be the computing corroded and expand all be extreme value computing, thus only burn into expansion image gradient information can be produced it is inevitable
Impact.
The content of the invention
For above-mentioned deficiency, the present invention proposes a kind of new image edge extraction method.
The purpose of the present invention is：Image border is extracted, makes the edge for extracting that there is more abundant detailed information, more
Integrity, seriality, multidirectional.
The technical scheme that adopted for achieving the above object of the present invention is：A kind of image edge extraction method, the method
Implementation steps it is as follows：
Step 1：Use low pass filter H0Inseparable additivity Multiscale Wavelet Decomposition is done to original image, low frequency is obtained
Figure and multilamellar high frequency subgraph.
Step 2：Morphological Gradient filter is carried out to the low frequency subgraph picture decomposited in the first step using morphological gradient
Ripple obtains filtered gradient figure.
Step 3：Multilamellar high frequency subgraph to obtaining in step one is added, and obtains enhancing the high frequency subgraph of marginal information
Picture, the modulus maximum of wavelet transformation is taken to high frequency subgraph, obtains a high frequency subgraph, i.e. edge graph.
Step 4：Gradient Fig. 1 and edge image 2 are done into additivity wavelet inverse transformation (addition), result gradient map is obtained.
Step 5：Binaryzation is carried out to the gradient map that the 4th step is obtained, preliminary edge figure is obtained.
Step 6：Using isolated point, deburring being removed in morphology and going H types point and edge unification etc. to operate, obtain
As a result edge graph.
The invention has the beneficial effects as follows：This method is simple to operate, Detection results are good；It can detect that gray value becomes in image
Change less target, obtain more edge details, can extract and there is integrity, multidirectional, continuously and translation invariance
Edge.Compared with other detection methods, this method also has processing speed faster.
Description of the drawings
Accompanying drawing is the flow chart of the present invention
Specific embodiment
This method constructs the inseparable low pass filter of two dimension；Multiple dimensioned inseparable additivity small echo is first done to original image
Decompose, the gradient map of the subimage after decomposing then is obtained using modern mathematics morphological gradient, high frequency subgraph is added
So as to extract low, high frequency subgraph edge respectively, final edge image is then reformed, and added using this two dimension is inseparable
Property small echo isotropism and Morphological Gradient detection edge it is abundant, quick the features such as, it is proposed that a kind of new image border carries
Take method.
Below in conjunction with flow chart, the present invention is described in detail.
First, low pass filter H0
It is isotropy because inseparable additivity small echo has good directivity, in order to obtain multidirectional image side
Edge information, the present invention selects the inseparable low pass filter of 6*6, and its building method is as follows：
First, it is [2,0 by flexible matrix；0,0], with compact schemes, to becoming second nature, the wave filter group of the 6*6 of orthogonality
It is expressed as：
(H0(x, y), H1(x, y), H2(x, y), H3(x, y))=T (1)
Wherein,UjOrthogonal Symmetric battle array centered on (j=1,2 ..., K), wiFor small echo plane,
D (x, y)=Diag (1, x, y, xy), V=(V0, V1, V2, V3) be orthogonal matrix, V1, V2, V3It is vectorial for 4 × 1, V0=(1,1,1,
1)T。
When K=2 is chosen, construction：
Formula (1) to formula (6) is wave filter, and original image obtains a low frequency subgraph and multilamellar high frequency after device after filtering
Image.
2nd, morphological gradient
In step 2, Morphological Gradient filtering is carried out to low frequency subgraph picture using morphological gradient, obtain filtering ladder
Degree figure, wherein Morphological Gradient is usually to describe one of gray-value variation situation in Image neighborhood by structural elements in mathematical morphology
Kind of gradient, in gray value morphology, most basic burn into dilation operation is respectively non-transform expansion and transform expansion, thus
Can produce with Gradient.
In the method, the following several grown form gradients of main construction：
1) expanded types：
2) corrosion type：T2=f-f Θ b
3) manipulation type is opened：
4) closed operation type：
Wherein,Expansive working is represented, Θ represents etching operation, and zero represents opening operation operation, ● represent closed operation operation.
Burn into expansion, open and close etc. operate the mathematical morphology that various ways can be formed by certain rational sequence combination
Gradient, in the method the Morphological Gradient of selection is in Image Edge-Detection：
First closed operation computing is carried out to original image, then dilation operation and opening operation are carried out respectively to result, finally asked
Both poor Morphological Gradient figures.
Expansion and corrosion are two kinds of most basic morphology operations, the combination of both computings during other computings, thus
This combination is the combination in certain sequence of maximum, minima in image in structural element limited range, and shape
State gradient is the difference of the value obtained by combining as two kinds, as long as and the infull phase of the gray value of structural element region
Together, then must there are maximum and minima.Herein in the method, the maximum of selection, the combinatorial operation mode of minima are distinguished
For：(bracket is suitable by from inside to outside for maximum (minima (maximum)) and maximum (minima (minima (maximum)))
Sequence computing), its Morphological Gradient for selecting is the difference of upper two kinds of compound modes, i.e. T=maximums (minima (maximum))-most
Big value (minima (minima (maximum))).
Clearly for the structural element window of arbitrary size, the average of above-mentioned expression formula is more than 0, therefore, it completely can be with
As gradient operator, and, only when the acquired results that first time dilation operation is made in structural element region are identical ash
During angle value, the value of above-mentioned expression formula is just 0, and the value in the case of other is both greater than 0, thus can well detect the change in image
Change.
4th, the acquisition of final edge image
Original image device after filtering, the multilamellar high frequency subgraph for obtaining is added the multilamellar high frequency subgraph for obtaining, this
When obtain enhancing the high frequency subgraph of marginal information, the modulus maximum of wavelet transformation is taken to high frequency subgraph, obtain one it is high
Frequency subimage, i.e. edge graph.Gradient map and edge image are done into additivity wavelet inverse transformation (addition), result gradient map is obtained.So
Afterwards the result gradient map to obtaining carries out binaryzation, obtains preliminary edge figure.Finally, using in morphology remove isolated point, go
Burr and go H types point and edge unification etc. operate, obtain result edge graph.
Claims (3)
1. a kind of image edge extraction method, the present invention relates to the neck such as target recognition, image segmentation, remote sensing, medical image analysis
Domain, is characterized in that：The step of the method, is as follows：
Step one：Use low pass filterInseparable additivity Multiscale Wavelet Decomposition is done to original image, low frequency is obtained
Figure and multilamellar high frequency subgraph,
Step 2：Morphological Gradient filtering is carried out to the low frequency subgraph picture decomposited in the first step using morphological gradient
Filtered gradient figure is obtained,
Step 3：Multilamellar high frequency subgraph to obtaining in step one is added, and obtains enhancing the high frequency subgraph of marginal information
Picture, the modulus maximum of wavelet transformation is taken to high frequency subgraph, obtains a high frequency subgraph, i.e. edge graph,
Step 4：Gradient map and edge image are done into additivity wavelet inverse transformation（It is added）, result gradient map is obtained,
Step 5：The gradient map obtained to step 4 carries out binaryzation, obtains preliminary edge figure,
Step 6：Using isolated point, deburring being removed in morphology and going H types point and edge unification etc. to operate, tied
Fruit edge graph.
2. a kind of image edge extraction method according to claim 1, is characterized in that：The foundation side of step 1 median filter
Method is as follows：
First, it is [2,0 by flexible matrix；0,0], with compact schemes, to becoming second nature, the wave filter group of the 6*6 of orthogonality represents
For：
（1）
（2）
Wherein,Centered on Orthogonal Symmetric battle array,For small echo
Plane,For orthogonal matrix,
For 41 is vectorial,
When K=2 is chosen, construction：
（3）
（4）
（5）
（6）.
3. a kind of image edge extraction method according to claim 1, is characterized in that：In step 2, morphological gradient
Algorithm it is as follows：
Several grown form gradients below main construction：
Wherein,Represent expansive working,Represent etching operation,Opening operation is operated,Represent closed operation operation
Burn into expansion, open and close etc. operate the Mathematical Morphology Gradient that various ways can be formed by certain rational sequence combination,
In the method the Morphological Gradient of selection is in Image Edge-Detection：
。
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610173046.2A CN106611424A (en) | 2016-03-23 | 2016-03-23 | Image edge extraction method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610173046.2A CN106611424A (en) | 2016-03-23 | 2016-03-23 | Image edge extraction method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106611424A true CN106611424A (en) | 2017-05-03 |
Family
ID=58614838
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610173046.2A Pending CN106611424A (en) | 2016-03-23 | 2016-03-23 | Image edge extraction method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106611424A (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108154507A (en) * | 2017-12-29 | 2018-06-12 | 长春师范大学 | Screwed pipe foreign matter detection system |
| CN108960180A (en) * | 2018-07-17 | 2018-12-07 | 泉州装备制造研究所 | Object detecting method based on the inseparable wavelet character of omnidirection two dimension |
| CN109712129A (en) * | 2018-12-25 | 2019-05-03 | 河北工业大学 | A kind of arc image processing method based on mathematical morphology |
| CN110954063A (en) * | 2018-09-27 | 2020-04-03 | 北京自动化控制设备研究所 | Optical relative measurement method for unmanned aerial vehicle landing recovery |
| CN111539312A (en) * | 2020-04-21 | 2020-08-14 | 罗嘉杰 | Method for extracting table from image |
| CN112184744A (en) * | 2020-11-29 | 2021-01-05 | 惠州高视科技有限公司 | Display screen edge defect detection method and device |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101930597A (en) * | 2010-08-10 | 2010-12-29 | 浙江大学 | Mathematical morphology-based image edge detection method |
-
2016
- 2016-03-23 CN CN201610173046.2A patent/CN106611424A/en active Pending
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101930597A (en) * | 2010-08-10 | 2010-12-29 | 浙江大学 | Mathematical morphology-based image edge detection method |
Non-Patent Citations (3)
| Title |
|---|
| 刘斌: "基于四通道不可分加性小波的多光谱图像融合", 《计算机学报》 * |
| 周树道: "基于多方向小波变换及形态学重构的ＳＡＲ图像边缘检测", 《解放军理工大学学报（自然科学版）》 * |
| 葛雯: "基于融合技术的小波变换和形态学边缘检测算法", 《东北大学学报(自然科学版)》 * |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108154507A (en) * | 2017-12-29 | 2018-06-12 | 长春师范大学 | Screwed pipe foreign matter detection system |
| CN108154507B (en) * | 2017-12-29 | 2023-06-13 | 长春师范大学 | Foreign matter detection system for threaded pipe |
| CN108960180A (en) * | 2018-07-17 | 2018-12-07 | 泉州装备制造研究所 | Object detecting method based on the inseparable wavelet character of omnidirection two dimension |
| CN110954063A (en) * | 2018-09-27 | 2020-04-03 | 北京自动化控制设备研究所 | Optical relative measurement method for unmanned aerial vehicle landing recovery |
| CN109712129A (en) * | 2018-12-25 | 2019-05-03 | 河北工业大学 | A kind of arc image processing method based on mathematical morphology |
| CN111539312A (en) * | 2020-04-21 | 2020-08-14 | 罗嘉杰 | Method for extracting table from image |
| CN112184744A (en) * | 2020-11-29 | 2021-01-05 | 惠州高视科技有限公司 | Display screen edge defect detection method and device |
| CN112184744B (en) * | 2020-11-29 | 2021-03-30 | 惠州高视科技有限公司 | Display screen edge defect detection method and device |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106611424A (en) | Image edge extraction method | |
| CN105046651B (en) | A kind of ultra-resolution ratio reconstructing method and device of image | |
| CN107358258B (en) | SAR image target classification based on NSCT double CNN channels and selective attention mechanism | |
| CN100573559C (en) | Self-adapting multi-dimension veins image segmenting method based on small echo and average drifting | |
| CN103208097B (en) | Filtering method is worked in coordination with in the principal component analysis of the multi-direction morphosis grouping of image | |
| US10275891B2 (en) | Phase transform for object and shape detection in digital images | |
| Bandyopadhyay | Pre-processing of mammogram images | |
| Lee et al. | Rotation symmetry group detection via frequency analysis of frieze-expansions | |
| CN101533507A (en) | Self-adaptive method for watermarking intensive image texture | |
| Mitra et al. | Intrinsic regularity detection in 3d geometry | |
| Liao et al. | Rotation-invariant texture features extraction using dual-tree complex wavelet transform | |
| Sayed et al. | Image object extraction based on curvelet transform | |
| CN105825514A (en) | Shear wave coefficient probability density distribution-based image edge detection method | |
| Zhang et al. | Texture feature-based image classification using wavelet package transform | |
| CN105069762A (en) | Image denoising method based on Shearlet transform and non-linear diffusion | |
| Mohebbian et al. | Increase the efficiency of DCT method for detection of copy-move forgery in complex and smooth images | |
| Gattim et al. | ROTATION AND SCALE INVARIANT FEATURE EXTRACTION FOR MRI BRAIN IMAGES. | |
| CN108090907B (en) | Image segmentation method based on non-Gaussian HMT model | |
| Ma | A revised edge detection algorithm based on wavelet transform for coal gangue image | |
| CN106327440B (en) | Picture breakdown filtering method containing non-local data fidelity term | |
| Zhao et al. | An improved Roberts edge detection algorithm based on mean filter and wavelet denoising | |
| CN104867149B (en) | Reproduction image authentication method based on part plan Linear Points | |
| Xiao | Second order morphology algorithm for IR small target enhancement in sea clutter background | |
| Wang et al. | Locally adaptive bivariate shrinkage algorithm for image denoising based on nonsubsampled contourlet transform | |
| Younan et al. | Edge detection in multispectral imagery via maximum entropy |
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 | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20170503 |