CN111369569A - Optimal scale parameter calculation method for multi-scale segmentation - Google Patents

Optimal scale parameter calculation method for multi-scale segmentation Download PDF

Info

Publication number
CN111369569A
CN111369569A CN202010111505.0A CN202010111505A CN111369569A CN 111369569 A CN111369569 A CN 111369569A CN 202010111505 A CN202010111505 A CN 202010111505A CN 111369569 A CN111369569 A CN 111369569A
Authority
CN
China
Prior art keywords
scale
segmentation
optimal
ground object
object target
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.)
Withdrawn
Application number
CN202010111505.0A
Other languages
Chinese (zh)
Inventor
王志华
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Institute of Geographic Sciences and Natural Resources of CAS
Original Assignee
Institute of Geographic Sciences and Natural Resources of CAS
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Institute of Geographic Sciences and Natural Resources of CAS filed Critical Institute of Geographic Sciences and Natural Resources of CAS
Priority to CN202010111505.0A priority Critical patent/CN111369569A/en
Publication of CN111369569A publication Critical patent/CN111369569A/en
Withdrawn legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/11—Region-based segmentation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/40—Analysis of texture
    • 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
    • G06T7/62—Analysis of geometric attributes of area, perimeter, diameter or volume
    • 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
    • G06T2207/10—Image acquisition modality
    • G06T2207/10032—Satellite or aerial image; Remote sensing

Abstract

The invention relates to 1, a multi-scale segmented optimal scale parameter calculation method, which is characterized by comprising the following steps: 1) the segmentation scale is set as a set S of a sequence according to exponential growth, the remote sensing image is segmented by the segmentation scale in the S, and the corresponding segmentation scale S is obtainediCorresponding number of divided objects Ni(ii) a 2) Using a segmentation scale SiAnd the number of divided objects NiFitting linear logarithm relation to obtain a segmentation scale SiAnd the number of divided objects NiThe slope K and the intercept C of the straight line after linear logarithmic relation fitting; 3) obtaining the number N of the segmentation objects of the ground object target according to the area A of the remote sensing image range and the average area a of the ground object targetaCalculating the optimal segmentation scale S of the ground object target according to the fitted straight linea. The method has the main characteristics that the optimal segmentation scale can be directly calculated according to the area of the ground object target to be analyzed, the manual visual segmentation result is avoided, and the automatic acquisition of the optimal segmentation scale parameters is realized.

Description

Optimal scale parameter calculation method for multi-scale segmentation
Technical Field
The invention belongs to the technical field of digital image processing, and particularly relates to an optimal scale parameter calculation method for multi-scale segmentation.
Background
Image segmentation is the basis for object-oriented classification. The conventional grid remote sensing image is converted into vector surface-shaped data in an image segmentation mode, so that the point, line and surface models in a classical geographic information system can be directly used for analyzing and interpreting the remote sensing image, and the accuracy and the category fineness of automatic interpretation of the remote sensing image are greatly improved. However, the optimal segmentation scale parameter setting in the currently widely used multi-scale segmentation algorithm has long been realized by manual intervention, and the automation degree of remote sensing image interpretation is seriously influenced. Because the segmentation scale parameters determine the size of the segmented object and determine the feature description of the segmented object, the accuracy of subsequent image interpretation is influenced, and therefore, the development of the optimal segmentation scale parameter setting research of multi-scale segmentation has important significance.
The existing optimal segmentation scale parameter setting mainly has three ideas: (1) and judging the segmentation result by manual visual observation, thereby selecting the corresponding optimal segmentation scale. Although the method has high precision, the method depends on the visual observation of human eyes and has no automation capability; (2) using the reference segmentation result, quantifying the similarity between the segmentation result with different scales and the reference segmentation result, thereby selecting an optimal segmentation scale (Su T, Zhang S]Isprs Journal of photometery and Remote Sensing,2017,130: 256-276). This type of method requires drawing of reference segmentation results, and this work still requires manual visual implementation. (3) Designing an unsupervised evaluation index, and quantitatively evaluating the segmentation result to select the optimal segmentation scale (
Figure BDA0002390170470000011
L,Csillik O,Eisank C,et al.Automated parameterisation for multi-scale image segmentationon multiple layers[J]ISPRS Journal of photographic and Remote Sensing,2014, 88: 119-127.). However, when the ground feature scene of the image is complex, the optimal segmentation scale selected by the method still needs manual visual check. It can be seen that the current three methods of thinking are all unable to be judged by manual visual observation, and seriously restrict the automation of remote sensing image interpretation.
According to the method, the optimal segmentation scale is calculated directly based on the area of the ground object target to be analyzed by constructing the data linear relation between the segmentation scale and the number of the segmented objects, and a key problem of remote sensing image interpretation automation is solved.
In the prior art, the method for directly calculating the optimal segmentation scale is not adopted for searching Chinese and foreign patent documents and the like.
Disclosure of Invention
Aiming at the problems, the invention provides an optimal scale parameter calculation method for multi-scale segmentation, which realizes the calculation of the optimal segmentation scale directly based on the area of a ground object target to be analyzed by constructing a data linear relation between the segmentation scale and the number of segmented objects.
The specific technical scheme of the invention is an optimal scale parameter calculation method for multi-scale segmentation, which is characterized by comprising the following steps of:
1) the division scale is set exponentially to a set S of series, see formula (I),
S＝{Si|Si＝TiS0,i＝0,1,2,…,n-1}……(I)
wherein S is0Is the starting segmentation scale, n is such that SnHas a number of divided objects of 1 but Sn-1Is a positive integer with the number of the divided objects larger than 1, T is a constant,
the remote sensing image is divided by the dividing scales to obtain corresponding dividing scales SiCorresponding number of divided objects Ni；
2) Using a segmentation scale SiAnd the number of divided objects NiFitting linear logarithm relation to obtain a segmentation scale SiAnd the number of divided objects NiThe slope K and the intercept C of the straight line after linear logarithmic relation fitting;
3) obtaining the number N of the segmentation objects of the ground object target according to the area A of the remote sensing image range and the average area a of the ground object targetaCalculating the optimal segmentation scale S of the ground object target according to the fitted straight linea。
Furthermore, the specific formula for fitting the log-linear relationship in step 2) is,
Figure BDA0002390170470000031
Figure BDA0002390170470000032
wherein x isiIs a division scale SiTaking the logarithmic value, yiIs the number of divided objects NiTaking the value after the logarithm of the number of the sample,
Figure BDA0002390170470000033
is the sample point xiThe average value of (a) of (b),
Figure BDA0002390170470000034
is the sample point yiIs measured.
Further, said step 1) is
Figure BDA0002390170470000035
S 01, x in said step 2)iIs a division scale SiTaking the value after base logarithm of 2, yiIs the number of divided objects NiThe base logarithm of 2 is taken.
Further, in the step 3), an optimal segmentation scale S of the ground object target is calculated according to the fitted straight lineaThe specific method of (3) is as follows,
Figure BDA0002390170470000036
the invention has the beneficial effects that: according to the method, the optimal segmentation scale is calculated directly based on the area of the ground object target to be analyzed by constructing the data linear relation between the segmentation scale and the number of the segmented objects, manual visual judgment is not needed, and the automation of the selection of the optimal segmentation scale is realized.
When the spectrum or the texture of the ground object target is relatively uniform and the size is relatively stable, the optimal segmentation scale can be directly calculated, and the method has great potential in the aspect of object-oriented automatic interpretation of the remote sensing image.
Drawings
FIG. 1 is a flow chart of the optimal scale parameter calculation of the present invention;
FIG. 2 is a diagram of an original remote sensing image used in one embodiment of the present invention;
FIG. 3 is a fitting relationship between a segmentation scale and a segmentation object data volume in an embodiment;
fig. 4 is a segmentation result corresponding to the optimal segmentation scale in the specific embodiment.
Detailed Description
The following describes the present invention with reference to the attached drawings.
In this embodiment, the remote sensing image processed by the method of the present invention is a SPOT 5 multispectral image and a panchromatic fused image in france, see fig. 2. the image spatial resolution is 2.5 meters, the image size is 400 rows and × 400 columns, and the image radiation resolution is 8 bits.
As shown in fig. 1, the method for calculating the optimal scale parameter of the multi-scale segmentation of the present invention comprises the following specific steps:
1) set the shape parameter 0.1, the compactness parameter 0.9, and then pair the segmentation scale with
Figure BDA0002390170470000041
Carrying out index sampling for the step length, carrying out remote sensing image segmentation by using the sampled segmentation scales, and obtaining S under the corresponding segmentation scalesiNumber of divided objects Ni. The set of segmentation scales obtained by exponential sampling is shown in formula (I),
Figure BDA0002390170470000042
wherein S is0Is the starting scale of the segmentation scale, and is set to 1 in this embodiment; n is such that SnHas a number of divided objects of 1 but Sn-1The number of the divided objects of (1) is a positive integer greater than 1, and is 16 in this embodiment.
2) Using a segmentation scale SiAnd the number of divided objects NiLinear of fitThe slope K and the intercept C in the logarithmic relation are specifically defined as,
Figure BDA0002390170470000043
Figure BDA0002390170470000044
wherein x isiIs a division scale SiTaking the value after base logarithm of 2, yiIs the number of divided objects NiTaking the value after taking the base logarithm of 2,
Figure BDA0002390170470000045
is the sample point xiThe average value of (a) of (b),
Figure BDA0002390170470000046
is the sample point yiIs measured. The fitting results are shown in FIG. 3.
3) Calculating the optimal segmentation scale of the ground object target according to the area A of the image range and the average area a of the ground object target, wherein the specific formula is as follows,
Figure BDA0002390170470000051
the above formula is obtained by dividing the ground object into the number of parts
Figure BDA0002390170470000052
Taking the logarithm with base 2 as
Figure BDA0002390170470000053
Substituted into formula (III) to obtain
Figure BDA0002390170470000054
After transformation of item shifting
Figure BDA0002390170470000055
Namely, it is
Figure BDA0002390170470000056
In this embodiment, a is 160000 pixels and a is 1200 pixels. The final calculated optimal segmentation scale is 41, and the corresponding segmentation result is shown in fig. 4.
According to the method, the optimal segmentation scale is calculated directly based on the area of the ground object target to be analyzed by constructing the data linear relation between the segmentation scale and the number of the segmented objects, manual visual judgment is not needed, and the automation of the selection of the optimal segmentation scale is realized. When the spectrum or the texture of the ground object target is relatively homogeneous and the size is relatively stable, the optimal segmentation scale can be directly calculated by the method, and the method has great potential in the aspect of object-oriented automatic interpretation of the remote sensing image.
It is to be understood that the embodiments described are only a few embodiments of the present invention, and not all embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.

Claims (4)

1. An optimal scale parameter calculation method for multi-scale segmentation is characterized by comprising the following steps:
1) the division scale is set exponentially to a set S of series, see formula (I),
S＝{Si|Si＝TiS0,i＝0,1,2,…,n-1}……(I)
wherein S is0Is the starting segmentation scale, n is such that SnHas a number of divided objects of 1 but Sn-1Is a positive integer with the number of the divided objects larger than 1, T is a constant,
the remote sensing image is divided by the dividing scale in the S to obtain the corresponding dividing scale SiCorresponding number of divided objects Ni；
2) Using a segmentation scale SiAnd the number of divided objects NiFitting linear logarithm relation to obtain a segmentation scale SiAnd the number of divided objects NiInclination of straight line after linear logarithmic relation fittingRate K and intercept C;
3) obtaining the number N of the segmentation objects of the ground object target according to the area A of the remote sensing image range and the average area a of the ground object targetaCalculating the optimal segmentation scale S of the ground object target according to the fitted straight linea。
2. The method for calculating the optimal scale parameter of the multi-scale segmentation as claimed in claim 1, wherein the specific formula for fitting the log-linear relationship in the step 2) is,
Figure FDA0002390170460000011
Figure FDA0002390170460000012
wherein x isiIs a division scale SiTaking the logarithmic value, yiIs the number of divided objects NiTaking the value after the logarithm of the number of the sample,
Figure FDA0002390170460000013
is the sample point xiThe average value of (a) of (b),
Figure FDA0002390170460000014
is the sample point yiIs measured.
3. The method as claimed in claim 2, wherein the optimal scale parameter calculation method in step 1) is
Figure FDA0002390170460000015
S01, x in said step 2)iIs a division scale SiTaking the value after base logarithm of 2, yiIs the number of divided objects NiThe base logarithm of 2 is taken.
4. Such as rightThe method for calculating the optimal scale parameter of the multi-scale segmentation according to claim 3, wherein the optimal segmentation scale S of the ground object target is calculated according to the fitted straight line in the step 3)aThe specific method of (3) is as follows,
Figure FDA0002390170460000021
CN202010111505.0A 2020-02-24 2020-02-24 Optimal scale parameter calculation method for multi-scale segmentation Withdrawn CN111369569A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010111505.0A CN111369569A (en) 2020-02-24 2020-02-24 Optimal scale parameter calculation method for multi-scale segmentation

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010111505.0A CN111369569A (en) 2020-02-24 2020-02-24 Optimal scale parameter calculation method for multi-scale segmentation

Publications (1)

Publication Number Publication Date
CN111369569A true CN111369569A (en) 2020-07-03

Family

ID=71209690

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010111505.0A Withdrawn CN111369569A (en) 2020-02-24 2020-02-24 Optimal scale parameter calculation method for multi-scale segmentation

Country Status (1)

Country Link
CN (1) CN111369569A (en)

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102880753A (en) * 2012-09-21 2013-01-16 武汉理工大学 Method for converting land utilization spatial characteristic scale based on fractal dimension
CN103646400A (en) * 2013-12-17 2014-03-19 中国地质大学（北京） Automatic scale segmentation parameter selection method for object remote sensing image analysis
CN105335965A (en) * 2015-09-29 2016-02-17 中国科学院遥感与数字地球研究所 High-resolution remote sensing image multi-scale self-adaptive decision fusion segmentation method
CN106651865A (en) * 2016-12-23 2017-05-10 湖北工业大学 Novel automatic selection method of optimal segmentation scale of high-resolution remote sensing image
CN110047079A (en) * 2019-04-26 2019-07-23 重庆交通大学 A kind of optimum segmentation scale selection method based on objects similarity

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102880753A (en) * 2012-09-21 2013-01-16 武汉理工大学 Method for converting land utilization spatial characteristic scale based on fractal dimension
CN103646400A (en) * 2013-12-17 2014-03-19 中国地质大学（北京） Automatic scale segmentation parameter selection method for object remote sensing image analysis
CN105335965A (en) * 2015-09-29 2016-02-17 中国科学院遥感与数字地球研究所 High-resolution remote sensing image multi-scale self-adaptive decision fusion segmentation method
CN106651865A (en) * 2016-12-23 2017-05-10 湖北工业大学 Novel automatic selection method of optimal segmentation scale of high-resolution remote sensing image
CN110047079A (en) * 2019-04-26 2019-07-23 重庆交通大学 A kind of optimum segmentation scale selection method based on objects similarity

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
ZHIHUA WANG, CHEN LU，XIAOMEI YANG: "Exponentially sampling scale parameters for the efficient segmentation of remote-sensing images", 《INTERNATIONAL JOURNAL OF REMOTE SENSING》 *
ZHIHUA WANG，XIAOMEI YANGA，CHEN LU，FENGSHUO YANG: "A scale self-adapting segmentation approach and knowledge transfer for automatically updating land use/cover change databases using high spatial resolution images", 《INT J APPL EARTH OBS GEOINFORMATION》 *
王志华，孟樊，杨晓梅，杨丰硕，方豫: "高空间分辨率遥感影像分割尺度参数自动选择研究", 《地球信息科学》 *

Similar Documents

Publication Publication Date Title
CN109063754B (en) Remote sensing image multi-feature joint classification method based on OpenStreetMap
CN107067405B (en) Remote sensing image segmentation method based on scale optimization
CN108428220B (en) Automatic geometric correction method for ocean island reef area of remote sensing image of geostationary orbit satellite sequence
CN112381013B (en) Urban vegetation inversion method and system based on high-resolution remote sensing image
Peng et al. Object-based change detection from satellite imagery by segmentation optimization and multi-features fusion
CN111968171A (en) Aircraft oil quantity measuring method and system based on artificial intelligence
CN110619263A (en) Hyperspectral remote sensing image anomaly detection method based on low-rank joint collaborative representation
CN111144250A (en) Land coverage classification method fusing radar and optical remote sensing data
CN116030352B (en) Long-time-sequence land utilization classification method integrating multi-scale segmentation and super-pixel segmentation
CN110826509A (en) Grassland fence information extraction system and method based on high-resolution remote sensing image
Li et al. Comparison of several remote sensing image classification methods based on envi
CN116012364B (en) SAR image change detection method and device
Zhu et al. Information extraction of high resolution remote sensing images based on the calculation of optimal segmentation parameters
CN114067118B (en) Processing method of aerial photogrammetry data
CN113936214B (en) Karst wetland vegetation community classification method based on fusion of aerospace remote sensing images
CN109063577B (en) Satellite image segmentation optimal segmentation scale determination method based on information gain rate
CN116863341B (en) Crop classification and identification method and system based on time sequence satellite remote sensing image
CN111882573A (en) Cultivated land plot extraction method and system based on high-resolution image data
CN112858181A (en) Black and odorous water body monitoring method and device and electronic equipment
CN115953604B (en) Real estate geographic information mapping data acquisition method
CN116994071A (en) Multispectral laser radar point cloud classification method based on self-adaptive spectrum residual error
CN111369569A (en) Optimal scale parameter calculation method for multi-scale segmentation
CN111666999A (en) Remote sensing image classification method
CN114742849B (en) Leveling instrument distance measuring method based on image enhancement
Zhao et al. Improving object-oriented land use/cover classification from high resolution imagery by spectral similarity-based post-classification

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
WW01 Invention patent application withdrawn after publication
WW01 Invention patent application withdrawn after publication

Application publication date: 20200703