CN106504244A - A kind of mine Edge-Detection Algorithm that is accumulated with wavelet scale based on Retinex - Google Patents
A kind of mine Edge-Detection Algorithm that is accumulated with wavelet scale based on Retinex Download PDFInfo
- Publication number
- CN106504244A CN106504244A CN201610951004.7A CN201610951004A CN106504244A CN 106504244 A CN106504244 A CN 106504244A CN 201610951004 A CN201610951004 A CN 201610951004A CN 106504244 A CN106504244 A CN 106504244A
- Authority
- CN
- China
- Prior art keywords
- image
- edge
- improved
- low
- wavelet
- 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
- 238000003708 edge detection Methods 0.000 title claims abstract description 24
- 238000000034 method Methods 0.000 claims abstract description 17
- 238000000354 decomposition reaction Methods 0.000 claims abstract description 15
- 238000001514 detection method Methods 0.000 claims abstract description 8
- 230000002708 enhancing effect Effects 0.000 claims description 12
- 230000009466 transformation Effects 0.000 claims description 7
- 101100001674 Emericella variicolor andI gene Proteins 0.000 claims description 6
- 238000006243 chemical reaction Methods 0.000 claims description 5
- 238000003384 imaging method Methods 0.000 claims description 5
- 238000010606 normalization Methods 0.000 claims description 2
- 238000001914 filtration Methods 0.000 abstract 1
- 230000000694 effects Effects 0.000 description 5
- 239000003245 coal Substances 0.000 description 4
- 238000012544 monitoring process Methods 0.000 description 4
- 238000004364 calculation method Methods 0.000 description 3
- 238000013461 design Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 238000004519 manufacturing process Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 2
- 238000013507 mapping Methods 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 101150070547 MAPT gene Proteins 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000008034 disappearance Effects 0.000 description 1
- 239000000428 dust Substances 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000005728 strengthening Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/70—Denoising; Smoothing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20048—Transform domain processing
- G06T2207/20064—Wavelet transform [DWT]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20172—Image enhancement details
- G06T2207/20192—Edge enhancement; Edge preservation
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Quality & Reliability (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Processing (AREA)
- Image Analysis (AREA)
Abstract
The present invention proposes a kind of mine Edge-Detection Algorithm based on Retinex and Wavelet Multiscale Product.One layer wavelet decomposition is carried out to input picture first, using the Multiscale Wavelet Decomposition algorithm process low frequency subgraph for having merged multiple dimensioned Retinex, two layers after obtaining filtering irradiation component and three layers of low frequency component, then improved fuzzy enhancement algorithm is used to high frequency subgraph, strengthen edge details while noise is suppressed, finally using the enhanced high frequency of improved multi-scale product edge detection algorithm process and low-frequency information, complete edge image is obtained.This paper algorithms have preferable accuracy of detection and accuracy, real-time, can meet the demand that pit robot is perceived for surrounding enviroment well, it is adaptable to the Image Edge-Detection under mine low light conditions.
Description
Technical field
Invention is related to a kind of mine Edge-Detection Algorithm that accumulates based on Retinex with wavelet scale, belongs to machine
Vision application..
Background technology
With the fast development of computer vision technique, video monitoring is with identification as raising coal production rate and automatization
The important means of the level of production, has gradually been applied in the safety in production in colliery.In Mine Monitoring cargo handling operation, will be varied
Robot introduce mine construction field, Mine Monitoring can be made more to rationalize, intelligent, such as：Digging robot, rescuing machine
Device people etc., they need to perceive surrounding enviroment by machine vision, provide important information for its control for walking work.Both at home and abroad
Research worker proposes multiple pit robot control systems based on video monitoring, and no matter which kind of system is required for acquisition
Borehole image is further processed, and the profile for extracting object in these processing procedures is the important step for carrying out object identification
Suddenly.
Traditional edge detection method adopts edge detection operator, such as Sobel, Prewitt, Laplace, Canny etc.,
Amount of calculation in detection is little, but noise resisting ability difference and easily causes the loss of effective detail edges.In recent years, with various new
Technology, the proposition of new theory, occur in that many new edge detection methods, typically have：It is based on artificial neural network, mathematics
The edge detection methods such as morphology, fuzzy theory, wavelet theory.But as underground coal mine bad environments, dust concentration are big, light
Not enough and uneven illumination so that conventional edge detection algorithm is difficult suitable for borehole image.And because underground coal mine operating mode special,
The research of borehole image edge detecting technology is still at an early stage at present.However, for ground low light conditions hypograph
Rim detection problem, existing more solution, for coal road Image Edge-Detection, a lot of technical thoughts also have reference or
Reference value.Wherein, because wavelet transformation can suppress noise under large scale, it is accurately positioned under little yardstick, therefore using multiple dimensioned
The fusion of small echo and other theories becomes the focus of such problem of solution, its advantage be with good local frequency domain characteristic and
Multiscale analysis ability, effectively can detect to abrupt local.And Retinex strengthens algorithm and can effectively improve figure
The visual effect of picture, improves its brightness uniformity, is shown the details being submerged in shadow region.
Content of the invention
A kind of present invention, it is proposed that the Edge-Detection Algorithm under low-light (level) environment suitable for mine.The algorithm exists
Be improved on the basis of Retinex image enhaucaments and multi-scale wavelet integration method, in high and low frequency component include information
The characteristics of, do different process respectively to subgraph.Removed in low frequency subgraph using improved multiple dimensioned Retinex (MSR) algorithm
Irradiation component, and using improved fuzzy enhancement algorithm process high frequency subgraph, made an uproar with strengthening the edge details of image and removing
Sound interference, finally obtains complete edge detail information using improved multi-scale product edge detection algorithm.New algorithm have compared with
Good edge continuity and accuracy, and the serious Image Edge-Detection of sound pollution can be solved the problems, such as well, it is adaptable to
Rim detection under the conditions of mine low-light (level) strong noise.
Description of the drawings
Accompanying drawing 1 is mine low-light (level) Edge-Detection Algorithm entire block diagram.
Accompanying drawing 2 is the multiple dimensioned Retinex algorithm for image enhancement of improved low frequency.
Specific embodiment
In order to deepen the understanding of the present invention, with reference to the accompanying drawings and examples invention is described further.
Fig. 1 is referred to, Fig. 1 is mine low-light (level) Edge-Detection Algorithm entire block diagram.The present invention is based on Retinex
Include that step is as follows with the mine Edge-Detection Algorithm of wavelet scale product.
S1:By original imageI(x,y) wavelet decomposition be four parts：I LL1(x,y)、I HL1(x,y)、I LH1(x,y)、I HH1
(x,y).
S2:Low frequency coefficient is continued to do two-layer wavelet decomposition and is removed in low frequency component using improved multiple dimensioned algorithm
Irradiation component.
Fig. 2 is referred to, Fig. 2 is the multiple dimensioned Retinex algorithm for image enhancement of improved low frequency.
This step adopts the multiple dimensioned Retinex algorithm for image enhancement of the low frequency based on three layers of wavelet decomposition, is returned first
One change is processed.
Due to Retinex algorithm process image range between [0,255], and through wavelet transformation after low-frequency information have
Just having negative, therefore to its do single scale conversion before, need to be by imageI LL1(x,y) be normalized, obtainI LL1 ’(x,y).
Computing formula is as follows：
Secondly using standard deviation it isσ 1 (1)Gaussian functionF 1 (1)(x,y) to normalized low frequency componentI LL1 ’(x,y) make single chi
Degree change, obtainsR LL1 (1)(x,y).
Computing formula is as follows：
.
Then willR LL1 (1)(x,y) be mapped to [I LL1(x,y)min,I LL1(x,y)max] in the range of, obtainR LL1 (1) ’(x,y),
Wavelet decomposition is carried out to which, is obtained：R LL1low (1)(x,y)、R LL1HL2 (1)(x,y)、R LL1LH2 (1)(x,y) andR LL1HH2 (1)(x,y).
In addition, the use of standard deviation being σ1 (2)Gaussian functionF 1 (2)(x,y) to low frequency componentI LL1 ’(x,y) do single scale change
Change, after mappingR LL1 (2) ’(x,y), after wavelet decomposition：R LL1low (2)(x,y)、R LL1HL2 (2)(x,y)、R LL1LH2 (2)(x,y) with
AndR LL1HH2 (2)(x,y).
Wherein, as high-frequency information reflects image detail, in order to obtain more detailed information, by wavelet decomposition after
High fdrequency components do following process.
.
Low frequency component contains the half-tone information of image, in order to preferably go back original image, as follows to low-frequency information process, its
Inω 1、ω 2For weights, andω 1+ω 2=1.
Finally, the low frequency component for upper step being obtainedI LL2(x,y) normalization obtainsI LL2 ’(x,y).Using standard deviation it isσ 2 (1)、σ 2 (2)Gaussian functionF 2 (1)(x,y)、F 2 (2)(x,y) right respectivelyI LL2 ’(x,y) single scale conversion twice is done, after mappingR LL2 (1) ’(x,y) withR LL2 (2) ’(x,y).Wavelet decomposition is carried out to above two amount, is respectively obtained：R LL2low (1)(x,y)、R LL2HL3 (1)(x,y)、R LL2LH3 (1)(x,y)、R LL2HH3 (1)(x,y) withR LL2low (2)(x,y)、R LL2HL3 (2)(x,y)、R LL2LH3 (2)(x,y)、R LL2HH3 (2)
(x,y).Three layers of wavelet low frequency and high fdrequency components are obtained after above-mentioned component fusionI LL3(x,y)、I HL3(x,y)、I LH3(x,y) andI HH3
(x,y).
The design is taken around function scale factorσ 1 (1)=2,σ 1 (2)=20；σ 2 (1)=2,σ 2 (2)=20, the now enhancing of image is imitated
Fruit is most preferably and without halation phenomenon；For the enhanced effect of balanced different parameters hypograph, weightsω 1、ω 2It is 0.5.
S3:To high frequency coefficientI HL1(x,y) andI LH1(x,y) do improved fuzzy enhancement algorithm to strengthen image edge thin
Save and remove noise jamming.
This step texture information larger to absolute value strengthens, and suppresses the less picture noise of absolute value.To being subordinate to
Degree function and enhancing function are improved, and are allowed to the enhancing for being more suitable for high frequency detail under low-light (level), and specific practice is as follows：
Using improved membership function, small echo high frequency imaging is transformed in fuzzy set；
Using improved enhancing function, fuzzy set is changed；
Image set is converted to enhanced fuzzy point set.
First, using improved membership function, small echo high frequency imaging is transformed in fuzzy set.According to image to be strengthened
Gamma characteristic, the versus grayscale grade from image is used as fuzzy characteristics.For suppressing the less portion of absolute value in high-frequency signal
Point, strengthen larger part, power function is chosen as membership function, reduce calculating on the premise of reinforced effects are ensured
Amount.
Expression formula is as follows：
.
Wherein,P ij For any pixel gradation of image；0<α<1, the curve shape of membership function is determined, should be according to concrete
Situation is chosen.In order to strengthen the Edge texture information in former high frequency imaging and suppress noise, the design takesλ=3.
Using improved enhancing function, fuzzy set is changed.Power law conversion can change image with logarithmic transformation
Dynamic range, but logarithmic function needs to can be only achieved the larger reinforced effects of amplitude through successive ignition, considerably increases calculation
The amount of calculation of method.And power law conversion is more flexibly, need to only change the size of power, just can reach different reinforced effects.
The expression formula of enhancing function is
.
The design takes membership function coefficientα=0.22.
Image set is converted to enhanced fuzzy point set.
The image in fuzzy characteristics domain is mapped in spatial domain using inverse transformation formula, is obtainedl ij , it is enhanced height
Frequency absolute coefficient.
Computing formula is：
.
Level detail image wavelet decompose after is processed respectively using improved fuzzy enhancement algorithmI HL1And vertical detailI LH1, obtain enhanced detailsI’ HL1WithI’ LH1.
S4:High frequency and the low frequency component that first two steps are obtained is processed using improved Wavelet Multiscale Product edge detection algorithm,
It is finally completed the rim detection of mine low-light (level) image.
Wavelet coefficient of the original image after high frequency and lower frequency region are processed under each yardstick is obtained in S2, S3 step.Take suitable
When wavelet coefficient weights, the multi-scale product on calculated level and vertical directionP x f(x,y) andP y f(x,y), and yardstick product
AmplitudePMf(x,y) and phase anglePAf(x,y).
Computing formula is：
.
Wherein,W 1 x f(x,y)、W 2 x f(x,y)、W 3 x f(x,y) it is three levels component；W 1 y f(x,y)、W 2 y f(x,y)、W 3 y f
(x,y) it is three layers of vertical component；λ1,λ2, λ3Weight for wavelet coefficient under different scale.Wavelet transformation system due to low yardstick
Number has higher precision, therefore selects suitable value to meet λ1>λ2>λ3, shared in multi-scale product to improve low multi-scale wavelet coefficient
Proportion.The present invention takes λ1=5, λ2=3, λ3=1.
WillPMf(x,y) alongPAf(x,y) modulus of local maximum point on direction, it is considered as candidate marginal.Along border
Marginal point under any yardstick is coupled together the modulus maxima curve that can be formed under the yardstick along border, the i.e. side of image in direction
Edge.
Through the edge image that multi-scale product is obtained, still contain a large amount of due to noise and the uneven falseness for causing of gray scale
Edge, in order to remove pseudo-edge, need to arrange corresponding threshold value.
Possible candidate edge pixel in image is tentatively extracted using auto-adaptive doublethreshold, then in the base of high threshold image
Edge is connected into profile on plinth, when running into breakpoint, is found in adjacent 8 points in the image that Low threshold is produced and be may be coupled to
The edge of image, like this, algorithm finds the junction point of the high threshold image point of interruption always in Low threshold image, until by height
Till threshold binary image is coupled together.
Due to having some short and small edges produced because of noise in the edge-detected image that obtains, now need to arrange length
Threshold value is removed it.Improved Wavelet Multiscale Product edge detection algorithm step is as follows.
Refined image edge, by the amplitude of each pixel and two neighbor amplitude contrasts along gradient direction, if
No more than neighbor, then the gradient zero setting；Otherwise it is taken as marginal point.
Calculate the auto-adaptive doublethreshold of each marginal point：Square region of the size as 10*10 is taken centered on each marginal point
Domain, in the region gradient rectangular histogram, taking percentage ratio isP 1The Grad at place is Low thresholdT low, taking percentage ratio isP 2The gradient at place
It is worth for high thresholdT high；WhereinP 1<P 2AndP 1,P 2∈[0% , 100% ].
It is less than in gradient mapT lowGradient magnitude zero setting, obtain imageG 1；It is less thanT highGradient magnitude zero setting, obtain figure
PictureG 2.
To imageG 2It is scanned, when running into first non-zero point A of gradient, starts with A as starting point, track with A as starting point
Contour line, until terminal B.
In imageG 1In find with B points in same position B ' points, if B ' point around 8 points in have non-zero pixels point
C ' is then in imageG 2In find corresponding C points and repeat previous step, tracking contour line is continued as starting point with C points, until two width figures
Till cannot continuing.This contour line of starting point for A is labeled as searching for.
Repeat above two steps, till it can not find new contour line, obtain edge image E0.
Length threshold is setL 0, in all contour lines that several steps above are found, length is less thanL 0Short and small edge go
Remove, obtain edge image E.So far, the rim detection of entire image is completed.
Percentage ratioP 1WithP 2Value have impact on final edge detection results, the details content of image withP 1、P 2Into
Instead, details content is less, easily causes the disappearance of image border, but excessive details content can be caused to do to follow-up feature identification
Disturb, consider pros and cons, the present invention takesP 1=20%,P 2=80%, the rim detection effect of image is most preferably.
Above content is to inventing made further description with reference to specific preferred implementation, it is impossible to assert this
Invention be embodied as be confined to these explanations.For the general technical staff of the technical field of the invention, not
On the premise of departing from present inventive concept, some simple deduction or replace can also be made, should all be considered as belonging to the guarantor of the present invention
Shield scope.
Claims (8)
1. a kind of mine Edge-Detection Algorithm based on Retinex and wavelet scale product, it is characterised in that including step
Suddenly:
It is four parts by original image wavelet decomposition：I LL1(x,y)、I HL1(x,y)、I LH1(x,y)、I HH1(x,y)；
To low frequency coefficientI LL1(x,y) continue to do two-layer wavelet decomposition and remove the photograph in low frequency component using improved MSR algorithms
Penetrate component；
To high frequency coefficientI HL1(x,y) andI LH1(x,y) do improved fuzzy enhancement algorithm to strengthen the edge details of image and go
Remove noise jamming；
High frequency and the low frequency component that first two steps are obtained is processed using improved Wavelet Multiscale Product edge detection algorithm, is finally completed
The rim detection of mine low-light (level) image.
2. according to claim 1 to low frequency coefficientI LL1(x,y) continue to do two-layer wavelet decomposition and calculated using improved MSR
Method removes the irradiation component in low frequency component, it is characterised in that
Using the multiple dimensioned Retinex algorithm for image enhancement of low frequency based on three layers of wavelet decomposition.
3. the multiple dimensioned Retinex algorithm for image enhancement of the low frequency based on three layers of wavelet decomposition is adopted according to claim 2, its
It is characterised by,
As the image range of Retinex algorithm process is between [0,255], and through wavelet transformation after low-frequency information have and just have
Negative, therefore to low frequency coefficientI LL1(x,y) do normalized；Gaussian function using various criterion difference is to normalization low frequency coefficient
Single scale conversion twice is done, is re-mapped in the range of corresponding coefficient of frequency, is then carried out wavelet decomposition respectively, obtain small echo low
Frequency and high fdrequency componentsI LL2(x,y)、I HL2(x,y)、I LH2(x,y) andI HH2(x,y)；Repeat said process, by three Wavelet Components
Three layers of wavelet low frequency and high fdrequency components are obtained after fusionI LL3(x,y)、I HL3(x,y)、I LH3(x,y) andI HH3(x,y).
4. according to claim 1 to high frequency coefficientI HL1(x,y) andI LH1(x,y) do improved fuzzy enhancement algorithm to strengthen
The edge details of image and remove noise do, it is characterised in that
Membership function and enhancing function are improved, the enhancing for being more suitable for high frequency detail under low-light (level) is allowed to.
5. according to claim 2 membership function and enhancing function are improved, are allowed to be more suitable under low-light (level)
The enhancing of high frequency detail, specific practice are as follows：
Using improved membership function, small echo high frequency imaging is transformed in fuzzy set；
Using improved enhancing function, fuzzy set is changed；
Image set is converted to enhanced fuzzy point set.
6. small echo high frequency imaging is transformed in fuzzy set using improved membership function according to claim 5, which is special
Levy and be,
According to the gamma characteristic of image to be strengthened, the versus grayscale grade from image is chosen power function and is made as fuzzy characteristics
For membership function.
7. fuzzy set is changed using improved enhancing function according to claim 5, it is characterised in that
The image in fuzzy characteristics domain is mapped in spatial domain using inverse transformation formula, obtains enhanced high frequency coefficient absolute
Value.
8. according to claim 1 using improved Wavelet Multiscale Product edge detection algorithm, it is characterised in that
Through the edge image that multi-scale product is obtained, possible candidate edge picture in image is tentatively extracted using auto-adaptive doublethreshold
Then edge is connected into profile on the basis of high threshold image by element, when running into breakpoint, is sought in the image that Low threshold is produced
Look for the edge that image is may be coupled in adjacent 8 points, like this, algorithm finds high threshold image always in Low threshold image
The junction point of the point of interruption, till high threshold image is coupled together.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610951004.7A CN106504244A (en) | 2016-11-03 | 2016-11-03 | A kind of mine Edge-Detection Algorithm that is accumulated with wavelet scale based on Retinex |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610951004.7A CN106504244A (en) | 2016-11-03 | 2016-11-03 | A kind of mine Edge-Detection Algorithm that is accumulated with wavelet scale based on Retinex |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106504244A true CN106504244A (en) | 2017-03-15 |
Family
ID=58323070
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610951004.7A Pending CN106504244A (en) | 2016-11-03 | 2016-11-03 | A kind of mine Edge-Detection Algorithm that is accumulated with wavelet scale based on Retinex |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106504244A (en) |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107330433A (en) * | 2017-05-17 | 2017-11-07 | 北京捷通华声科技股份有限公司 | Image processing method and device |
| CN108389214A (en) * | 2018-03-06 | 2018-08-10 | 青岛海信医疗设备股份有限公司 | The processing method and processing device of ultrasonoscopy, electronic equipment, storage medium |
| CN108563986A (en) * | 2018-03-02 | 2018-09-21 | 中国人民武装警察部队总医院 | Earthquake region electric pole posture judgment method based on wide-long shot image and system |
| CN109358063A (en) * | 2018-04-22 | 2019-02-19 | 杨娟 | It hurdles the detection method of integrity degree |
| CN109919960A (en) * | 2019-02-22 | 2019-06-21 | 西安工程大学 | A kind of image continuous boundary detection method based on Multiscale Gabor Filters device |
| CN110197476A (en) * | 2019-04-24 | 2019-09-03 | 武汉科技大学 | A kind of analysis method of the three-dimensional micro- mine phase of the complicated sinter based on Fusion Features |
| CN112037248A (en) * | 2020-01-03 | 2020-12-04 | 哈尔滨理工大学 | Mine underground personnel tracking algorithm based on computer vision |
| CN113223642A (en) * | 2021-04-30 | 2021-08-06 | 中国矿业大学 | Simulation method for detecting toxic gas by cobalt-doped modified indium nitride sensor |
| CN113989128A (en) * | 2020-07-31 | 2022-01-28 | 山东财经大学 | Based on improved L0Image smoothing method and system of gradient minimization model |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102289670A (en) * | 2011-08-31 | 2011-12-21 | 长安大学 | Image characteristic extraction method with illumination robustness |
| CN102682436A (en) * | 2012-05-14 | 2012-09-19 | 陈军 | Image enhancement method on basis of improved multi-scale Retinex theory |
| CN105844601A (en) * | 2016-05-20 | 2016-08-10 | 中国矿业大学(北京) | Mine image enhancement method based on bilateral filtering and multi-scale Retinex algorithm |
-
2016
- 2016-11-03 CN CN201610951004.7A patent/CN106504244A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102289670A (en) * | 2011-08-31 | 2011-12-21 | 长安大学 | Image characteristic extraction method with illumination robustness |
| CN102682436A (en) * | 2012-05-14 | 2012-09-19 | 陈军 | Image enhancement method on basis of improved multi-scale Retinex theory |
| CN105844601A (en) * | 2016-05-20 | 2016-08-10 | 中国矿业大学(北京) | Mine image enhancement method based on bilateral filtering and multi-scale Retinex algorithm |
Non-Patent Citations (4)
| Title |
|---|
| 应东杰等: ""煤矿监控图像增强算法的分析与实现"", 《工矿自动化》 * |
| 曹万鹏等: ""一种照明无关的小波尺度相乘边缘检测方法"", 《光学学报》 * |
| 汪荣贵等: ""基于小波域信息融合的MSR改进算法"", 《中国图象图形学报》 * |
| 贾超等: ""利用小波多尺度积实现裂纹缺陷的边缘检测"", 《计算机工程与应用》 * |
Cited By (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107330433B (en) * | 2017-05-17 | 2020-01-07 | 北京捷通华声科技股份有限公司 | Image processing method and device |
| CN107330433A (en) * | 2017-05-17 | 2017-11-07 | 北京捷通华声科技股份有限公司 | Image processing method and device |
| CN108563986A (en) * | 2018-03-02 | 2018-09-21 | 中国人民武装警察部队总医院 | Earthquake region electric pole posture judgment method based on wide-long shot image and system |
| CN108563986B (en) * | 2018-03-02 | 2024-04-05 | 中国人民武装警察部队总医院 | Method and system for judging posture of telegraph pole in jolt area based on long-distance shooting image |
| CN108389214A (en) * | 2018-03-06 | 2018-08-10 | 青岛海信医疗设备股份有限公司 | The processing method and processing device of ultrasonoscopy, electronic equipment, storage medium |
| CN109358063A (en) * | 2018-04-22 | 2019-02-19 | 杨娟 | It hurdles the detection method of integrity degree |
| CN109358063B (en) * | 2018-04-22 | 2019-10-01 | 河北金特力斯体育设施有限公司 | It hurdles the detection method of integrity degree |
| CN109919960B (en) * | 2019-02-22 | 2023-04-07 | 西安工程大学 | Image continuous edge detection method based on multi-scale Gabor filter |
| CN109919960A (en) * | 2019-02-22 | 2019-06-21 | 西安工程大学 | A kind of image continuous boundary detection method based on Multiscale Gabor Filters device |
| CN110197476A (en) * | 2019-04-24 | 2019-09-03 | 武汉科技大学 | A kind of analysis method of the three-dimensional micro- mine phase of the complicated sinter based on Fusion Features |
| CN110197476B (en) * | 2019-04-24 | 2021-04-23 | 武汉科技大学 | Analysis method of complex sinter three-dimensional micro-mineral phase based on feature fusion |
| CN112037248A (en) * | 2020-01-03 | 2020-12-04 | 哈尔滨理工大学 | Mine underground personnel tracking algorithm based on computer vision |
| CN113989128B (en) * | 2020-07-31 | 2022-07-26 | 山东财经大学 | Based on improved L 0 Image smoothing method and system of gradient minimization model |
| CN113989128A (en) * | 2020-07-31 | 2022-01-28 | 山东财经大学 | Based on improved L0Image smoothing method and system of gradient minimization model |
| CN113223642B (en) * | 2021-04-30 | 2023-09-15 | 中国矿业大学 | Simulation method for detecting toxic gas by cobalt doped modified indium nitride sensor |
| CN113223642A (en) * | 2021-04-30 | 2021-08-06 | 中国矿业大学 | Simulation method for detecting toxic gas by cobalt-doped modified indium nitride sensor |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106504244A (en) | A kind of mine Edge-Detection Algorithm that is accumulated with wavelet scale based on Retinex | |
| Ying et al. | Beamlet transform‐based technique for pavement crack detection and classification | |
| CN105046677B (en) | A kind of enhancing treating method and apparatus for traffic video image | |
| CN103136766B (en) | A kind of object conspicuousness detection method based on color contrast and color distribution | |
| Xu et al. | Pavement crack detection based on saliency and statistical features | |
| CN101777181B (en) | Ridgelet bi-frame system-based SAR image airfield runway extraction method | |
| CN107274365A (en) | A kind of mine image intensification method based on unsharp masking and NSCT algorithms | |
| CN108564814A (en) | A kind of parking position detection method and device based on image | |
| CN107633491A (en) | A kind of area image Enhancement Method and storage medium based on target detection | |
| WO2022141145A1 (en) | Object-oriented high-resolution remote sensing image multi-scale segmentation method and system | |
| CN103279957A (en) | Method for extracting remote sensing image interesting area based on multi-scale feature fusion | |
| CN104376551A (en) | Color image segmentation method integrating region growth and edge detection | |
| CN106846289A (en) | A kind of infrared light intensity and polarization image fusion method based on conspicuousness migration with details classification | |
| CN104036465A (en) | Edge detection based on self-adaptive nonlocal mean denoising method | |
| CN108564597A (en) | A kind of video foreground target extraction method of fusion gauss hybrid models and H-S optical flow methods | |
| Ma et al. | The Canny edge detection and its improvement | |
| Zhang et al. | Single image dehazing based on fast wavelet transform with weighted image fusion | |
| CN107729814A (en) | A kind of method and device for detecting lane line | |
| CN111652033A (en) | Lane line detection method based on OpenCV | |
| Wu et al. | Research on rail surface defect detection method based on UAV images | |
| CN106384338A (en) | Enhancement method for light field depth image based on morphology | |
| Jia et al. | Research on the decomposition and fusion method for the infrared and visible images based on the guided image filtering and Gaussian filter | |
| CN113592750A (en) | Infrared enhancement method based on gradient histogram | |
| Li et al. | A study of crack detection algorithm | |
| Wu et al. | Research on crack detection algorithm of asphalt pavement |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | 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 |
Application publication date: 20170315 |
|
| WD01 | Invention patent application deemed withdrawn after publication |