CN106778774B - High-resolution remote sensing image artificial ground feature contour detection method - Google Patents
High-resolution remote sensing image artificial ground feature contour detection method Download PDFInfo
- Publication number
- CN106778774B CN106778774B CN201611052418.2A CN201611052418A CN106778774B CN 106778774 B CN106778774 B CN 106778774B CN 201611052418 A CN201611052418 A CN 201611052418A CN 106778774 B CN106778774 B CN 106778774B
- Authority
- CN
- China
- Prior art keywords
- remote sensing
- sensing image
- contour
- curve
- detecting
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The invention relates to a method for detecting the contour of an artificial ground object of a high-resolution remote sensing image. The method comprises the following steps: step 1, firstly, input remote sensing images are inputIDown-sampling to obtainI s Will beI s Convolving with Gaussian kernel function to obtain imageI sc Will beI s MinusI sc Obtaining a difference imageD(ii) a Step 2, comparing the difference images by using a gray statistic comparison windowDPerforming edge detection to obtain a candidate curve set; step 3, verifying and screening the candidate curve set by adopting a standard error function-based verification method to obtain a carefully selected curve set; and 4, smoothing the selected curve set, and outputting a high-resolution remote sensing image artificial ground feature contour detection result. The method and the device furthest excavate the contour information of the artificial ground object in the remote sensing image, and can be applied to accurate extraction of the artificial ground objects such as buildings, roads and the like.
Description
Technical Field
The invention relates to the field of remote sensing image processing, in particular to a method for detecting an artificial ground feature contour of a high-resolution remote sensing image.
Background
The remote sensing image is the comprehensive reflection of the spectrum and the geometric characteristics of the ground target and the phenomenon on the image, not only is a pixel unit set which embodies the brightness characteristic and the chromaticity characteristic, but also has complex spectrum characteristic and structure characteristic. The high-resolution remote sensing image is a very special digital image, and the complexity degree of the high-resolution remote sensing image is far higher than that of a common image. Contour detection of high-resolution remote sensing images is the basis of analysis and understanding of remote sensing information and is a challenge in the field of digital image processing. At present, many scholars continuously propose relevant outline detection theories and methods, but in view of published results, the methods have the following problems:
(1) because the remote sensing image is influenced by a plurality of factors such as a sensor, the position of the sun and the like in the imaging process, the target ground object information represented in the image is not only incomplete, but also contains a large amount of noise. The edge and noise are both shown as large abrupt change of gray scale in the spatial domain, and are reflected as high-frequency components in the frequency domain, and the noise is often detected as an edge point as a result of performing contour detection on the edge and the true contour is not detected due to noise interference.
(2) The remote sensing image has abundant information, and compared with a common image, the remote sensing image contains much more content, the mutual influence and interference of information among different ground objects and an insignificant or fuzzy boundary make it very difficult to extract an interested target contour.
Disclosure of Invention
The invention provides a method for detecting the contour of an artificial ground object of a high-resolution remote sensing image, which takes the high-resolution remote sensing image as a data source and makes a difference image on a scale space, can reduce the interference of noise on the extracted contour and furthest reserve the edge information of the image, and has the advantages of small calculated amount, strong self-adaptive capacity and reliable output result.
The technical scheme adopted for realizing the aim of the invention is as follows: the method comprises the following steps:
step 1: firstly, the input remote sensing image I is sampled down to obtain IsIs shown bysConvolving with a Gaussian kernel function to obtain an image IscIs shown bysSubtract IscObtaining a difference image D;
step 2: carrying out edge detection on the difference image D in the step 1 by using a gray scale statistic comparison window GSC to obtain a candidate curve set Scont;
And step 3: considering the rule that the gray scale statistic comparison window GSC in the step 2 obeys normal distribution, a standard error function-based verification method is adopted to perform verification on the candidate curve set S in the step 2contAll the curves in (1) are verified and screened to obtain a selected curve set Fcont;
And 4, step 4: for the selected curve set F of step 3contAnd smoothing all the curves in the image to output a high-resolution remote sensing image artificial ground feature contour detection result.
Standard deviation of the Gaussian kernel functionWhere k is a trade-off coefficient used to balance the jaggies and blur degree of the profile and s is the ratio of adjacent scales in scale space.
The gray scale statistic comparison window GSC is movable, and the moving mode adopts a progressive scanning mode.
The gray scale statistic comparison window GSC is composed of a bright template MLAnd a dark template MDAnd the size of an area contained in the window is determined by the width w of the side surface of the window formed by the adjacent windows, and the value v of the candidate curve in the gray statistic comparison window GSC is calculated according to the following formula:
in the formula (1), M and N are respectively bright templates MLAnd a dark template MDThe number of the pixel points included in (a),Lmand DnRespectively being bright templates MLM-th pixel in (1) and dark template MDWhen the gray scale statistic comparison window GSC does not contain boundary information, the nth pixel covers a homogeneous area, LnIs constantly equal to DnI.e. delta (L)m,Dn) When the content is equal to 0.5,when the gray scale statistical comparison window GSC contains significant boundary information, LnIs constantly less than DnI.e. delta (L)m,Dn) When 1, v is M × N; when the value v of the candidate curve is larger than the threshold value T1And taking the candidate curve as a refining curve.
The side width w is a bright template MLAnd a dark template MDThe short side of the minimum external rectangle adopts two valuesWherein the content of the first and second substances,guarantee bright template MLAnd a dark template MDAt least one row of pixels is covered for detecting a high contrast contour.Guarantee bright template MLAnd a dark template MDAt least three rows of pixels are covered for detecting the contour containing noise and fuzzy edge.
The standard error function-based verification method is characterized in that a standard error value SER of a curve is calculated by using the following formula:
in the formula (2), the upper limit of integrationIn the expression of hWhen the standard error value SER of the refined curve is less than the threshold value T2When it is, it is taken as the selection curve.
The smoothing method adopts a Bezier curve smoothing method.
The candidate curve is a result obtained by directly detecting with an edge detection algorithm, wherein the result contains an excessively short or invalid edge, and further verification and screening are required in subsequent steps.
The invention has the beneficial effects that: the method and the device furthest excavate the contour information of the artificial ground object in the remote sensing image, and can be applied to accurate extraction of the artificial ground objects such as buildings, roads and the like.
Drawings
FIG. 1 is an overall process flow diagram of the present invention.
Fig. 2 is a schematic diagram of a gray scale statistical comparison window according to the present invention.
Detailed Description
The following detailed description of embodiments of the invention refers to the accompanying drawings.
FIG. 1 is an overall process flow diagram of the present invention:
in step 101, the input high-resolution remote sensing image I to be processed is a Quickbird full-color image with a spatial resolution of 0.81 m.
In step 102, the high-resolution remote sensing image I is down-sampled to obtain IsWhere the sampling interval is 2 pixels.
In step 103, I is addedsConvolving with a Gaussian kernel function to obtain an image IscIs shown bysSubtract IscObtaining a difference image D in which the standard deviation of the Gaussian kernel functionWhere k is a trade-off coefficient used to balance the jaggies and blur degree of the profile and s is the ratio of adjacent scales in scale space. After repeated experiments, k was set to 1.5 and s was set to 0.8.
In step 104, the difference image D in step 103 is subjected to edge detection by using the gray scale statistical comparison window GSC to obtain a candidate curve set ScontThe GSC is movable, the movement mode adopts a progressive scanning mode, and the artificial ground features based on the high-resolution remote sensing image follow the characteristics of regular distributionConsidering the balance between the calculated amount of the subsequent steps and the retention degree of the profile related information, performing tests by using remote sensing images with different sensors and different sizes, finding that the effect of setting the moving step pitch P to be 5 is the best, and setting the gray scale statistic comparison window GSC by using a bright template MLAnd a dark template MDAnd the size of an area contained in the window is determined by the width w of the side surface of the window formed by the adjacent windows, and the value v of the candidate curve in the gray statistic comparison window GSC is calculated according to the following formula:
wherein M and N are respectively bright templates MLAnd a dark template MDThe number of the pixel points included in (a),Lmand DnRespectively being bright templates MLM-th pixel in (1) and dark template MDWhen the gray scale statistic comparison window GSC does not contain boundary information, the n-th pixel covers a homogeneous region without boundary information, and LnIs constantly equal to DnI.e. delta (L)m,Dn) 0.5) of the total weight of the composition,when the gray scale statistical comparison window GSC contains significant boundary information, LnIs constantly less than DnI.e. delta (L)m,Dn) 1), v is mxn; when the value v of the candidate curve is larger than the threshold value T1And taking the candidate curve as a refining curve. The width w of the side surface adopts two valuesWherein the content of the first and second substances,guarantee bright template MLAnd a dark template MDAt least one row of pixels is covered for detecting a high-contrast contour;guarantee bright template MLAnd a dark template MDAt least three rows of pixels are covered for detecting the contour containing noise and fuzzy edge.
In step 105, considering the rule that the gray scale statistic comparison window GSC in step 104 obeys normal distribution, a standard error function-based verification method is adopted to verify the candidate curve set S in step 4contAll the curves in (1) are verified and screened to obtain a selected curve set FcontThe standard error function-based verification method comprises the following steps of calculating a standard error value SER of a curve by using the following formula:
wherein, the upper limit of integrationIn the expression of hWhen the standard error value SER of the refined curve is less than the threshold value T2When it is, it is taken as the selection curve.
At step 106, the selected curve set F of step 105 is processedcontAll the curves in (1) are smoothed by a bezier curve smoothing method.
In step 107, the profile is output.
Upon testing, the threshold T in step 1041And a threshold T in step 1062Set to 0.75 and 0.6, respectively.
Fig. 2 is a schematic diagram of a gray scale statistical comparison window according to the present invention.
200 is a part of the difference image D, 201 is a section of contour curve, 202 is a bright template MLAnd 203 is a dark template MDAnd 204 is the side width w.
Claims (7)
1. A method for detecting the contour of an artificial ground object of a high-resolution remote sensing image is characterized by comprising the following steps:
step 1: firstly, the input remote sensing image I is sampled down to obtain IsIs shown bysConvolving with a Gaussian kernel function to obtain an image IscIs shown bysSubtract IscObtaining a difference image D;
step 2: carrying out edge detection on the difference image D in the step 1 by using a gray scale statistic comparison window GSC to obtain a candidate curve set Scont;
And step 3: considering the rule that the gray scale statistic comparison window GSC in the step 2 obeys normal distribution, a standard error function-based verification method is adopted to perform verification on the candidate curve set S in the step 2contAll the curves in (1) are verified and screened to obtain a selected curve set Fcont;
And 4, step 4: for the selected curve set F of step 3contAll the curves in the image are respectively smoothed, and a high-resolution remote sensing image artificial ground feature contour detection result is output;
the gray scale statistic comparison window GSC is composed of a bright template MLAnd a dark template MDAnd the size of an area contained in the window is determined by the width w of the side surface of the window formed by the adjacent windows, and the value v of the candidate curve in the gray statistic comparison window GSC is calculated according to the following formula:
in the formula (1), M and N are respectively bright templates MLAnd a dark template MDThe number of the pixel points included in (a),Lmand DnRespectively being bright templates MLM-th pixel in (1) and dark template MDThe nth pixel in (1); when the gray scale statistic comparison window GSC does not contain boundary information, covering a homogeneous region, LmIs constantly equal to DnI.e. delta (L)m,Dn) 0.5) of the total weight of the composition,when the gray scale statistical comparison window GSC contains significant boundary information, LmIs constantly less than DnI.e. delta (L)m,Dn) When 1, v is M × N; when the value v of the candidate curve is larger than the threshold value T1And taking the candidate curve as a refining curve.
3. The method for detecting the artificial terrain contour of the high-resolution remote sensing image according to claim 1, wherein the gray scale statistic comparison window GSC is movable, and the moving mode adopts a line-by-line scanning mode.
4. The method for detecting the contour of the artificial ground object of the remote sensing image with high resolution as claimed in claim 1, wherein the width w of the side surface is a bright template MLAnd a dark template MDThe short side of the minimum external rectangle adopts two valuesWherein the content of the first and second substances,guarantee bright template MLAnd a dark template MDAt least one row of pixels is covered for detecting a high-contrast contour;guarantee bright template MLAnd a dark template MDCovering at least three rows of pixels for detecting noise-containing, edge-obscuringA contour.
5. The method for detecting the artificial terrain contour of the high-resolution remote sensing image according to claim 1, wherein the standard error function-based verification method is to calculate the standard error value SER of the curve by using the following formula:
6. The method for detecting the artificial terrain contour of the high-resolution remote sensing image according to claim 1, characterized in that the smoothing method adopts a Bezier curve smoothing method.
7. The method for detecting the artificial terrain contour of the high-resolution remote sensing image as claimed in claim 1, wherein the candidate curve is a result obtained by directly detecting with an edge detection algorithm, and contains an excessively short or invalid edge, which requires further verification and screening by subsequent steps.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201611052418.2A CN106778774B (en) | 2016-11-25 | 2016-11-25 | High-resolution remote sensing image artificial ground feature contour detection method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201611052418.2A CN106778774B (en) | 2016-11-25 | 2016-11-25 | High-resolution remote sensing image artificial ground feature contour detection method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN106778774A CN106778774A (en) | 2017-05-31 |
CN106778774B true CN106778774B (en) | 2020-04-03 |
Family
ID=58912363
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201611052418.2A Active CN106778774B (en) | 2016-11-25 | 2016-11-25 | High-resolution remote sensing image artificial ground feature contour detection method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106778774B (en) |
Families Citing this family (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111047616B (en) * | 2019-12-10 | 2021-03-19 | 中国人民解放军陆军勤务学院 | Remote sensing image landslide target constraint active contour feature extraction method |
CN113313724B (en) * | 2021-05-27 | 2022-04-08 | 深圳企业云科技股份有限公司 | Line detection processing method for resisting resampling of mobile phone camera |
Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101777181A (en) * | 2010-01-15 | 2010-07-14 | 西安电子科技大学 | Ridgelet bi-frame system-based SAR image airfield runway extraction method |
CN102253184A (en) * | 2011-06-29 | 2011-11-23 | 南京信息工程大学 | Remote sensing inversion method for land surface evapotranspiration of arid and semi-arid regions |
CN103208011A (en) * | 2013-05-05 | 2013-07-17 | 西安电子科技大学 | Hyperspectral image space-spectral domain classification method based on mean value drifting and group sparse coding |
CN103559493A (en) * | 2013-10-21 | 2014-02-05 | 中国农业大学 | Method for extracting linear ground objects |
CN103839265A (en) * | 2014-02-26 | 2014-06-04 | 西安电子科技大学 | SAR image registration method based on SIFT and normalized mutual information |
CN103886597A (en) * | 2014-03-24 | 2014-06-25 | 武汉力成伟业科技有限公司 | Circle detection method based on edge detection and fitted curve clustering |
CN104794495A (en) * | 2015-05-04 | 2015-07-22 | 福建师范大学 | Large-format remote-sensing image region classifying method based on straight line statistical characteristics |
CN105740873A (en) * | 2016-02-01 | 2016-07-06 | 福建师范大学 | Artificial feature straight line contour detection method of remote-sensing image |
-
2016
- 2016-11-25 CN CN201611052418.2A patent/CN106778774B/en active Active
Patent Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101777181A (en) * | 2010-01-15 | 2010-07-14 | 西安电子科技大学 | Ridgelet bi-frame system-based SAR image airfield runway extraction method |
CN102253184A (en) * | 2011-06-29 | 2011-11-23 | 南京信息工程大学 | Remote sensing inversion method for land surface evapotranspiration of arid and semi-arid regions |
CN103208011A (en) * | 2013-05-05 | 2013-07-17 | 西安电子科技大学 | Hyperspectral image space-spectral domain classification method based on mean value drifting and group sparse coding |
CN103559493A (en) * | 2013-10-21 | 2014-02-05 | 中国农业大学 | Method for extracting linear ground objects |
CN103839265A (en) * | 2014-02-26 | 2014-06-04 | 西安电子科技大学 | SAR image registration method based on SIFT and normalized mutual information |
CN103886597A (en) * | 2014-03-24 | 2014-06-25 | 武汉力成伟业科技有限公司 | Circle detection method based on edge detection and fitted curve clustering |
CN104794495A (en) * | 2015-05-04 | 2015-07-22 | 福建师范大学 | Large-format remote-sensing image region classifying method based on straight line statistical characteristics |
CN105740873A (en) * | 2016-02-01 | 2016-07-06 | 福建师范大学 | Artificial feature straight line contour detection method of remote-sensing image |
Non-Patent Citations (7)
Title |
---|
ICA-ASIFT-BASED multi-temporal matching of high-resolution remote sensing urban images;fenghua huang等;《cybernetics and information technologies》;20161020;第16卷(第5期);第1629-1637页 * |
methods and example for remote sensing data assimilation in land surface process modeling;H.banch等;《IEEE TRANSACTIONS ON GEOSCIENCE AND REOMOTE SENSING》;20030818;第41卷(第7期);第34-49页 * |
一种基于灰度差统计的边缘检测方法;殷润民等;《计算机工程》;20060420;第32卷(第8期);第201-203页 * |
基于灰度的图像边缘检测与匹配算法的研究;傅立光等;《电脑知识与技术》;20160818;第12卷(第17期);第175-176页 * |
基于边缘像元投影的微小轴承亚像素边缘检测;徐平等;《仪器仪表学报》;20161015;第37卷(第10期);第2300-2306页 * |
海上溢油遥感图像的边缘检测算法研究;景雨;《中国博士学位论文全文数据库 信息科技辑》;20120515(第05期);I140-65 * |
自适应多方向灰度形态学图像边缘检测算法;曹风云等;《光学技术》;20160515;第42卷(第3期);第234-238页 * |
Also Published As
Publication number | Publication date |
---|---|
CN106778774A (en) | 2017-05-31 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108629775B (en) | Thermal state high-speed wire rod surface image processing method | |
CN109242888B (en) | Infrared and visible light image fusion method combining image significance and non-subsampled contourlet transformation | |
CN104834912B (en) | A kind of weather recognition methods and device based on image information detection | |
CN109345508A (en) | A kind of Assessing Standards For Skeletal method based on two stages neural network | |
CN109993797B (en) | Door and window position detection method and device | |
CN109242870A (en) | A kind of sea horizon detection method divided based on image with textural characteristics | |
CN106204617B (en) | Adapting to image binarization method based on residual image histogram cyclic shift | |
JP2006285310A (en) | Evaluation method of canopy of forest, and its canopy evaluation program | |
CN108961307B (en) | Circular target scale self-adaptive tracking method based on nuclear correlation filtering tracking | |
CN108830808B (en) | On-satellite infrared image stripe noise removing method based on similar line window mean value compensation | |
Pan et al. | No-reference assessment on haze for remote-sensing images | |
CN114926407A (en) | Steel surface defect detection system based on deep learning | |
CN110889843A (en) | SAR image ship target detection method based on maximum stable extremal region | |
CN114549441A (en) | Sucker defect detection method based on image processing | |
CN106778774B (en) | High-resolution remote sensing image artificial ground feature contour detection method | |
CN117036341A (en) | Pavement crack detection method based on image processing | |
Fang et al. | Image quality assessment on image haze removal | |
CN104637060B (en) | A kind of image partition method based on neighborhood principal component analysis-Laplce | |
CN105279244B (en) | The method for building up in sugarcane seed bud characteristic classification data library | |
CN107369163B (en) | Rapid SAR image target detection method based on optimal entropy dual-threshold segmentation | |
CN109035228A (en) | A kind of radioscopic image processing method of non-uniform thickness component | |
KR101426864B1 (en) | Region-based edge enhancement method for short-distance thermal target tracking | |
CN117011291A (en) | Watch shell quality visual detection method | |
CN112288726A (en) | Method for detecting foreign matters on belt surface of underground belt conveyor | |
CN112258495A (en) | Building wood crack identification method based on convolutional neural network |
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 | ||
GR01 | Patent grant | ||
GR01 | Patent grant |