CN108053439A - A kind of computational methods of the river gravel psephicity based on image procossing - Google Patents
A kind of computational methods of the river gravel psephicity based on image procossing Download PDFInfo
- Publication number
- CN108053439A CN108053439A CN201711064715.3A CN201711064715A CN108053439A CN 108053439 A CN108053439 A CN 108053439A CN 201711064715 A CN201711064715 A CN 201711064715A CN 108053439 A CN108053439 A CN 108053439A
- Authority
- CN
- China
- Prior art keywords
- point
- gradient
- edge
- pixel
- focus
- 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
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/60—Analysis of geometric attributes
-
- 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
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30181—Earth observation
- G06T2207/30184—Infrastructure
Landscapes
- Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Geometry (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a kind of computational methods of the river gravel psephicity based on image procossing, ellipse fitting is carried out to rock edge using least square method after improvement, the criterion of a logical oval similarity is given based on obtained data, the quantitative round and smooth degree for analyzing rock, compared with the mode of direct visual perception, with more reliability, repeatability and perspective.
Description
Technical field
The invention belongs to geologic survey Exploration Domain, more particularly to a kind of river gravel psephicity based on image procossing
Computational methods.
Background technology
In geological exploration field, the shape of river gravel is the basic parameter of gravel, closer to the degree of circle, illustrates rock
Stone be handled upside down away from discrete time.Gravel illustrates that rock is handled upside down longer away from discrete time closer to circle.Geologist uses
Psephicity describes it.Circularity is a mathematical concept, and earliest definition refers to most sharp arm of angle radius of curvature and most long diameter half
Ratio (Wentworth C K 1919), concept is modified to the average song on all angles or side in particle by Wadell H (1932)
Rate radius and the radius ratio received in maximum.Due to the complexity of geological survey, generally use and elliptical similarity degree at present
Rock and elliptical similarity degree are estimated as the parameter of assessment.
However in actual mechanical process, traditional mathematical concept is simultaneously not easy to field operation, and usual researcher is logical
Cross the round and smooth degree that direct visual perception judges rock.By the feature of its external form, round as a ball shape, round shape, secondary circle are divided into
Shape, subangular, angular, cusped edge horn shape carry out the arrangement of related data and analysis with this.But such dividing mode mark
Accurate very fuzzy, subjectivity is too strong, therefore such as how a kind of quantitative mode divides rock, becomes one and urgently solves
Certainly the problem of.
The content of the invention
For too subjective only to visually observe the method for the round and smooth degree of rock at present, reliability is low, and gravel is carried
The problem of judgement of history is too assumed.The present invention proposes a kind of new method of the solution gravel circularity based on image procossing,
Accurate description not only has been carried out to the round and smooth degree of river bed rock, but also has given the quantitative criteria to round and smooth degree, has been phase
It closes research and provides effective help.
The present invention is to propose in order to solve the above problem, it is therefore an objective in geologic survey field, to river gravel psephicity
Judge to provide a quantitative criteria being conveniently operated.
In the present invention, to achieve these goals, first, edge is carried out to river gravel image based on Canny operators
Extraction selects two-dimensional Gaussian function to filter out noise signal, obtains smoothed image.Two-dimensional Gaussian function：
Wherein σ is standard value, and for controlling smoothness, (x, y) is the coordinate of pixel in image.
Amplitude and the direction of the gradient of each pixel of river gravel image are calculated, the gradient G of horizontal direction is obtainedx(x,y)
With the gradient G of vertical directiony(x, y) and gradient G ' (x, y)：
Gx(x, y)=G (x+1, y)-G (x, y)
Gy(x, y)=G (x, y+1)-G (x, y)
Non-maxima suppression traversing graph picture is carried out, if the gray value of pixel and the gray scale of front and rear two pixels on its gradient direction
For value compared to not being the largest, pixel value is set to 0, i.e., is not edge.Since pixel is discrete Two-Dimensional Moment in digital picture
Battle array, the point of gradient direction both sides are not necessarily present, therefore the gray value of this point needs to obtain by the point interpolation of both sides.Interpolation
Method is as follows：
When the gradient magnitude in x directions is larger, weight is represented with weight1, when the gradient magnitude in y directions is larger, power
Weight2 is reused to represent.The Grad Grad_1 and Grad_2 put on gradient direction by neighborhood graded elemental value according to weight into
Row interpolation obtains.
Weight1=abs (Gy(x,y)/Gx(x,y))
Weight2=abs (Gx(x,y)/Gy(x,y))
Grad_1=weight1 × Grad1+ (1-weight1) × Grad2
Grad_2=weight2 × Grad3+ (1-weight2) × Grad4
Wherein Grad1, Grad2, Grad3, Grad4 divide following four situation discussion.
When X-direction amplitude is larger, it is divided into following two situations：
(3)Gx(x,y)×Gy(x, y) ＞ 0,
Grad1=G'(x+1, y+1)
Grad2=G'(x+1, y)
Grad3=G'(x-1, y-1)
Grad4=G'(x-1, y)
(4)Gx(x,y)×Gy(x, y) ＜ 0,
Grad1=G'(x+1, y-1)
Grad2=G'(x+1, y)
Grad3=G'(x-1, y+1)
Grad4=G'(x-1, y)
When Y-direction amplitude is larger, it is divided into following two situations：
(3)Gx(x,y)×Gy(x, y) ＞ 0
Grad1=G'(x-1, y-1)
Grad2=G'(x, y-1)
Grad3=G'(x+1, y+1)
Grad4=G'(x, y+1)
(4)Gx(x,y)×Gy(x, y) ＜ 0
Grad1=G'(x+1, y-1)
Grad2=G'(x, y-1)
Grad3=G'(x-1, y+1)
Grad4=G'(x, y+1)
After acquiring gradient, carry out dual threshold and determine edge and connection edge.For each pixel, if its gradient is higher than height
Threshold value then regards as edge, less than Low threshold, then regards as non-edge.It is positioned there between then according to around the pixel eight
Whether a point has really for the point of marginal point, if so, it is marginal point then to regard as the point, otherwise, regards as non-edge point.
After edge extracting completion, the ellipse fitting based on least square method is carried out.The midpoint of edge data is calculated, it will
Apart from the position of each 10 pixel units in midpoint or so as initial focus group, by each focus and around it eight points combination
Referred to as focus sequence.Any point in each focus sequence can be combined as one group with any point in another focus sequence
81 groups of focal groups can be combined out per a pair of initial focus group in focal group.Each marginal point is calculated each focal group respectively to two
Range data obtained by each focal group is carried out mean square error calculating by the sum of distance of a focus, finds out mean square error minimum
Focal group.When the focal group of mean square error minimum is initial focus group, it is believed that ellipse fitting is completed, the focus
The as most suitable elliptical focus of combination, using the average value of the sum of the distance of most suitable elliptic focus and edge each point as most suitable ellipse
Long axial length, and thereby determine that most suitable ellipse.And calculate most suitable elliptic parameter.
After ellipse fitting is completed, oval similarity is defined, i.e., the business of mean square error and most suitable ellipse diameter is as river
The quantitative criteria of gravel psephicity.The root mean square of wherein a diameter of most suitable elliptical major and minor axis.
Compared with prior art, the present invention advantageous effect is：The present invention is using least square method after improving to rock side
Edge carries out ellipse fitting, and the criterion of a logical oval similarity, quantitative analysis are given based on obtained data
The round and smooth degree of rock, compared with the mode of direct visual perception, with more reliability, repeatability and perspective.
Description of the drawings
Fig. 1 is the flow chart of the entire algorithm of the present invention.
Specific embodiment
A kind of computational methods of river gravel psephicity based on image procossing proposed by the present invention, below in conjunction with the accompanying drawings 1
It is as follows that the present invention is described in more detail.
The implementation of the present invention is mainly in two sub-sections.Edge extraction techniques based on Canny operators are with being based on minimum
The ellipse fitting of square law method.Edge extraction techniques include step 1~4,11,12, and ellipse fitting technology includes step 5~10.
Step 1：The image pixel value of gravel image is switched into gray value, two-dimensional Gaussian function is selected to carry out gaussian filtering,
Form smoothed image.Two-dimensional Gaussian function：
Wherein σ is standard value, and for controlling smoothness, (x, y) is the coordinate of pixel in image.
Step 2：The data obtained after gaussian filtering are handled.Amplitude and the direction of gradient are calculated, level side is obtained
To gradient GxThe gradient G of (x, y) and vertical directiony(x, y) and gradient G ' (x, y)：
Gx(x, y)=G (x+1, y)-G (x, y)
Gy(x, y)=G (x, y+1)-G (x, y)
Step 3：Carry out non-maxima suppression.Traversing graph picture, if the gray value of pixel and front and rear two pictures on its gradient direction
For the gray value of element compared to not being the largest, pixel value is set to 0, i.e., is not edge.Since pixel is discrete in digital picture
Two-dimensional matrix, the point of gradient direction both sides are not necessarily present, therefore the gray value of this point needs to obtain by the point interpolation of both sides
It arrives.Interpolation method is as follows：
When the gradient magnitude in x directions is larger, weight is represented with weight1, when the gradient magnitude in y directions is larger, power
Weight2 is reused to represent.The Grad Grad_1 and Grad_2 put on gradient direction by neighborhood graded elemental value according to weight into
Row interpolation obtains.
Weight1=abs (Gy(x,y)/Gx(x,y))
Weight2=abs (Gx(x,y)/Gy(x,y))
Grad_1=weight1 × Grad1+ (1-weight1) × Grad2
Grad_2=weight2 × Grad3+ (1-weight2) × Grad4
Wherein Grad1, Grad2, Grad3, Grad4 divide following four situation discussion.
When X-direction amplitude is larger, it is divided into following two situations：
(1)Gx(x,y)×Gy(x, y) ＞ 0,
Grad1=G'(x+1, y+1)
Grad2=G'(x+1, y)
Grad3=G'(x-1, y-1)
Grad4=G'(x-1, y)
(2)Gx(x,y)×Gy(x, y) ＜ 0,
Grad1=G (x+1, y-1)
Grad2=G'(x+1, y)
Grad3=G'(x-1, y+1)
Grad4=G'(x-1, y)
When Y-direction amplitude is larger, it is divided into following two situations：
(1)Gx(x,y)×Gy(x, y) ＞ 0
Grad1=G'(x-1, y-1)
Grad2=G'(x, y-1)
Grad3=G'(x+1, y+1)
Grad4=G'(x, y+1)
(2)Gx(x,y)×Gy(x, y) ＜ 0
Grad1=G'(x+1, y-1)
Grad2=G'(x, y-1)
Grad3=G'(x-1, y+1)
Grad4=G'(x, y+1)
Step 4：It determines initial dual threshold, carries out dual threashold value filtering and determine edge and connection edge.For each pixel,
If its gradient is higher than high threshold, edge is regarded as, less than Low threshold, then regards as non-edge.Then root positioned there between
Whether have really for the point of marginal point according to eight points around the pixel, if so, it is marginal point then to regard as the point, otherwise, regard as
Non-edge point.
Step 5：The edge data extracted is done to the early-stage preparations being fitted.Preliminary preparation has：It is by Data Integration
One two-dimensional array calculates the point midway of edge data, using apart from the position of each 10 pixel units in midpoint or so as just
Beginning focal group, by each focus and around it combination of eight points be known as focus sequence.
Step 6：Totally 81 each the combining for focus combination formed to two focus sequences calculate each edge respectively
O'clock the sum of to the distance of two focuses, range data carries out mean square error calculating obtained by each putting combination, finds out mean square error
Poor minimum point combination, if the point of mean square error minimum is combined as initial focus group, jumps to step 9, otherwise, carries out step 7.
Step 7：Initial focus group is moved at the point combination of mean square error minimum, re-establishes focus sequence.
Step 8：Judge that whether initial focus moves number more than 300 times, if so, jumping to step 11.Otherwise, rebound step
6。
Step 9：Ellipse fitting is completed, calculate most suitable elliptical parameter and is recorded.
Step 10：Most suitable oval and edge mean square error is calculated with elliptical diameter ratio, for describing river gravel
With elliptical similarity degree, wherein diameter is defined as transverse and the root mean square of short axle.Flow terminates.
Step 11：High threshold in dual threshold increases 5 certainly.
Step 12：Judge whether high threshold is more than 50.If so, carry out step 13.Otherwise, return to step 4.
Step 13：Excessively high in threshold value, Xun Huan 300 times or more can not still be fitted success, illustrate that target data is very few
Or noise signal is strong, can not filter out totally, explanation can not fitted ellipse.Flow terminates.
Claims (3)
1. a kind of computational methods of the river gravel psephicity based on image procossing, which is characterized in that comprise the following steps：
Step 1) is based on Canny operators and carries out edge extracting to river gravel image, and two-dimensional Gaussian function is selected to filter out noise letter
Number, obtain smoothed image, two-dimensional Gaussian function：
Wherein σ is standard value, and for controlling smoothness, (x, y) is the coordinate of pixel in image；
Step 2) calculates amplitude and the direction of the gradient of each pixel of river gravel image, and the gradient G of horizontal direction is obtainedx(x,y)
With the gradient G of vertical directiony(x, y) and gradient G ' (x, y)：
Gx(x, y)=G (x+1, y)-G (x, y)
Gy(x, y)=G (x, y+1)-G (x, y)
Step 3) carries out non-maxima suppression traversing graph picture, if the gray value of pixel and the ash of front and rear two pixels on its gradient direction
Angle value is not compared to being the largest, and pixel value is set to 0, i.e., is not edge, since pixel is discrete Two-Dimensional Moment in digital picture
Battle array, the point of gradient direction both sides are not necessarily present, therefore the gray value of this point needs to obtain by the point interpolation of both sides；
After step 4) acquires gradient, carry out dual threshold and determine edge and connection edge.For each pixel, if its gradient is higher than
High threshold then regards as edge, less than Low threshold, then regards as non-edge, it is positioned there between then according to the pixel around
Whether eight points have really for the point of marginal point, if so, it is marginal point then to regard as the point, otherwise, regard as non-edge point；
Step 5) carries out the ellipse fitting based on least square method, finds out most suitable ellipse, and calculate after edge extracting completion
Most suitable elliptic parameter；
Step 6) defines oval similarity, i.e., the business of mean square error and most suitable ellipse diameter is as river after ellipse fitting is completed
The quantitative criteria of gravel psephicity is flowed, wherein the root mean square of a diameter of most suitable elliptical major and minor axis.
2. a kind of computational methods of river gravel psephicity based on image procossing according to claim 1, feature exist
In interpolation method is as follows in step 3)：
When the gradient magnitude in x directions is larger, weight is represented with weight1, and when the gradient magnitude in y directions is larger, weight is used
Weight2 represents that the Grad Grad_1 and Grad_2 put on gradient direction is inserted by neighborhood graded elemental value according to weight
It is worth to；
Weight1=abs (Gy(x,y)/Gx(x,y))
Weight2=abs (Gx(x,y)/Gy(x,y))
Grad_1=weight1 × Grad1+ (1-weight1) × Grad2
Grad_2=weight2 × Grad3+ (1-weight2) × Grad4
Wherein Grad1, Grad2, Grad3, Grad4 divide following four situation discussion；
When X-direction amplitude is larger, it is divided into following two situations：
(1)Gx(x,y)×Gy(x, y) ＞ 0,
Grad1=G'(x+1, y+1)
Grad2=G'(x+1, y)
Grad3=G'(x-1, y-1)
Grad4=G'(x-1, y)
(2)Gx(x,y)×Gy(x, y) ＜ 0,
Grad1=G'(x+1, y-1)
Grad2=G'(x+1, y)
Grad3=G'(x-1, y+1)
Grad4=G'(x-1, y)
When Y-direction amplitude is larger, it is divided into following two situations：
(1)Gx(x,y)×Gy(x, y) ＞ 0
Grad1=G'(x-1, y-1)
Grad2=G'(x, y-1)
Grad3=G'(x+1, y+1)
Grad4=G'(x, y+1)
(2)Gx(x,y)×Gy(x, y) ＜ 0
Grad1=G'(x+1, y-1)
Grad2=G'(x, y-1)
Grad3=G'(x-1, y+1)
Grad4=G'(x, y+1).
3. a kind of computational methods of river gravel psephicity based on image procossing according to claim 1, feature exist
In the ellipse fitting method based on least square method is as follows in step 5)：
Calculate edge data midpoint, using apart from the position of each 10 pixel units in midpoint or so as initial focus group, will be every
A focus and around it combination of eight points be known as focus sequence, any point in each focus sequence can be with another focus
Any point in sequence is combined as one group of focal group, 81 groups of focal groups can be combined out per a pair of initial focus group, to each coke
Point group calculates each marginal point the sum of to the distance of two focuses respectively, range data obtained by each focal group is carried out square
Error calculation finds out the focal group of mean square error minimum, and if only if the focal group of mean square error minimum be initial focus group
When, it is believed that ellipse fitting is completed, which combines as most suitable elliptical focus, with most suitable elliptic focus and edge each point away from
From the sum of average value be used as most suitable elliptical long axial length, and thereby determine that most suitable ellipse.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711064715.3A CN108053439A (en) | 2017-11-02 | 2017-11-02 | A kind of computational methods of the river gravel psephicity based on image procossing |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711064715.3A CN108053439A (en) | 2017-11-02 | 2017-11-02 | A kind of computational methods of the river gravel psephicity based on image procossing |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108053439A true CN108053439A (en) | 2018-05-18 |
Family
ID=62118889
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711064715.3A Pending CN108053439A (en) | 2017-11-02 | 2017-11-02 | A kind of computational methods of the river gravel psephicity based on image procossing |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108053439A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110716239A (en) * | 2018-07-11 | 2020-01-21 | 中石化石油工程技术服务有限公司 | Fine evaluation method for lithology of well logging gravel rock mass based on electrical imaging |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2006350557A (en) * | 2005-06-14 | 2006-12-28 | Canon Inc | Image processor and its method |
| KR101592333B1 (en) * | 2014-09-30 | 2016-02-05 | 국방과학연구소 | Apparatus and method for detecting ellipse in images of optical satellite |
| CN106250687A (en) * | 2016-07-27 | 2016-12-21 | 长江大学 | Go the deposit gravel roundness calculation method of fasciation IPP |
| CN107292900A (en) * | 2017-07-05 | 2017-10-24 | 黄河科技学院 | A kind of method for detecting image edge and device based on Canny algorithms |
-
2017
- 2017-11-02 CN CN201711064715.3A patent/CN108053439A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2006350557A (en) * | 2005-06-14 | 2006-12-28 | Canon Inc | Image processor and its method |
| KR101592333B1 (en) * | 2014-09-30 | 2016-02-05 | 국방과학연구소 | Apparatus and method for detecting ellipse in images of optical satellite |
| CN106250687A (en) * | 2016-07-27 | 2016-12-21 | 长江大学 | Go the deposit gravel roundness calculation method of fasciation IPP |
| CN107292900A (en) * | 2017-07-05 | 2017-10-24 | 黄河科技学院 | A kind of method for detecting image edge and device based on Canny algorithms |
Non-Patent Citations (2)
| Title |
|---|
| 王红义: "砾石磨圆度椭圆拟合及其地质意义探讨", 《2017中国地球科学联合学术年会论文集（十六）》 * |
| 金刚: "自适应Canny算法研究及其在图像边缘检测中的应用", 《中国优秀硕士论文全文数据库科技信息辑》 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110716239A (en) * | 2018-07-11 | 2020-01-21 | 中石化石油工程技术服务有限公司 | Fine evaluation method for lithology of well logging gravel rock mass based on electrical imaging |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106447669B (en) | The adhesion particle image concave point dividing method differentiated based on round masking-out area ratio | |
| CN110349207B (en) | Visual positioning method in complex environment | |
| CN108597009B (en) | Method for detecting three-dimensional target based on direction angle information | |
| CN104299008B (en) | Vehicle type classification method based on multi-feature fusion | |
| CN108876799B (en) | Real-time step detection method based on binocular camera | |
| CN104992429B (en) | One kind is based on the enhanced mountain cracks detection method of image local | |
| CN105469094B (en) | A kind of edge vectors line drawing method of road surface bianry image | |
| CN111739064B (en) | Method for tracking target in video, storage device and control device | |
| CN107862735B (en) | RGBD three-dimensional scene reconstruction method based on structural information | |
| CN106204570B (en) | A kind of angular-point detection method based on non-causal fractional order gradient operator | |
| CN108010075B (en) | Local stereo matching method based on multi-feature combination | |
| CN106504289B (en) | indoor target detection method and device | |
| CN106530313A (en) | Sea-sky line real-time detection method based on region segmentation | |
| CN108921813A (en) | Unmanned aerial vehicle detection bridge structure crack identification method based on machine vision | |
| CN110992288B (en) | Video image blind denoising method used in mine shaft environment | |
| CN110414385A (en) | A kind of method for detecting lane lines and system based on homography conversion and characteristic window | |
| CN105043395B (en) | A kind of real-time Dynamic Location method of aircraft menology soft landing | |
| CN109886935A (en) | A kind of road face foreign matter detecting method based on deep learning | |
| CN108564620A (en) | Scene depth estimation method for light field array camera | |
| CN109102013A (en) | A kind of improvement FREAK Feature Points Matching digital image stabilization method suitable for tunnel environment characteristic | |
| CN111080696A (en) | Underwater sea cucumber identification and positioning method based on computer vision | |
| CN108898148A (en) | A kind of digital picture angular-point detection method, system and computer readable storage medium | |
| CN103617631B (en) | A kind of tracking based on Spot detection | |
| CN110245600A (en) | Adaptively originate quick stroke width unmanned plane Approach for road detection | |
| CN105787912A (en) | Classification-based step type edge sub pixel localization method |
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 |
Application publication date: 20180518 |
|
| WD01 | Invention patent application deemed withdrawn after publication |