CN111402279B - SAR image edge feature extraction method based on combined filter - Google Patents

SAR image edge feature extraction method based on combined filter Download PDF

Info

Publication number
CN111402279B
CN111402279B CN202010053326.6A CN202010053326A CN111402279B CN 111402279 B CN111402279 B CN 111402279B CN 202010053326 A CN202010053326 A CN 202010053326A CN 111402279 B CN111402279 B CN 111402279B
Authority
CN
China
Prior art keywords
straight line
operator
filtering
image
value
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.)
Expired - Fee Related
Application number
CN202010053326.6A
Other languages
Chinese (zh)
Other versions
CN111402279A (en
Inventor
王涛
窦利军
张艳
芮军栋
张永生
于英
江刚武
王康康
李磊
李力
王龙辉
周丽雅
张正豪
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.)
Information Engineering University of PLA Strategic Support Force
Original Assignee
Information Engineering University of PLA Strategic Support Force
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 Information Engineering University of PLA Strategic Support Force filed Critical Information Engineering University of PLA Strategic Support Force
Priority to CN202010053326.6A priority Critical patent/CN111402279B/en
Publication of CN111402279A publication Critical patent/CN111402279A/en
Application granted granted Critical
Publication of CN111402279B publication Critical patent/CN111402279B/en
Expired - Fee Related legal-status Critical Current
Anticipated expiration legal-status Critical

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
    • G06T7/13—Edge detection
    • 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
    • G06T2207/10044—Radar image
    • 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
    • G06T2207/20024—Filtering details
    • 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
    • G06T2207/20212—Image combination
    • G06T2207/20221—Image fusion; Image merging

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Image Processing (AREA)
  • Radar Systems Or Details Thereof (AREA)

Abstract

The invention belongs to the technical field of image processing, and particularly relates to an SAR image edge feature extraction method based on a combined filter. According to the invention, the Gabor filtering operator and the Gauss gamma-shaped double-window filtering operator are combined to construct a combined filter operator according to the good multi-scale characteristic and multi-directional characteristic of the Gabor filtering operator, and the constructed combined filter operator is utilized to filter the image to obtain the edge characteristic of the image so as to obtain the image after edge extraction. The image can eliminate false response generated by factors such as noise and the like, can also improve the processing effect on a high reflection region in the SAR image, and has a good edge detection effect.

Description

SAR image edge feature extraction method based on combined filter
Technical Field
The invention belongs to the technical field of image processing, and particularly relates to an SAR image edge feature extraction method based on a combined filter.
Background
Synthetic Aperture Radar (SAR) is a high-resolution microwave side-looking radar, the SAR adopts a range projection mode for imaging, the purpose of imaging is achieved by recording a time sequence of received backscattering waves, the structural information is rich, the imaging is not limited by time or affected by weather, and the SAR has wide application.
The trihedron formed by ground targets such as buildings has strong reflection characteristics on radar signals, and very obvious edge structure characteristics are presented on radar images. The earliest edge detection method for the SAR image uses an edge detection method for the optical image, such as a filter operator like Canny, Sobel and the like, which is influenced by imaging characteristics, so that strong speckle noise exists on the SAR image.
In the prior art, a wavelet transform and median filtering combined method is usually adopted, and a rough contour of an image can be extracted by obtaining high-frequency information of the image, but false responses are more, the contour is incomplete, and information loss is serious. In order to solve the problem, a ratio-based filter operator, such as a Gaussian-Gamma-Shaped Bi-window filter operator (GGS-Bi), may be used, since in the process of extracting the edge structure of the SAR image, the contribution of points closer to the edge of the SAR image is larger, the GGS-Bi window is designed such that the image point weight of adjacent edge points decreases from the center point to the outside, and the edge strength of the image can be reflected more truly, and the window uses a Gaussian function along the edge direction and a Gamma function perpendicular to the edge direction. Generally, the filter operator is sensitive to a high reflection region, has a poor effect on a low reflection region, and has low tolerance to noise. However, the speckle noise in the high reflection region in the SAR image has a large influence, so that the filtering operator has a poor processing effect on the high reflection region and a poor edge detection effect.
Disclosure of Invention
The invention provides an SAR image edge feature extraction method based on a combined filter, which is used for solving the problem of poor edge detection effect caused by a filter operator only based on a ratio.
In order to solve the technical problem, the technical scheme of the invention comprises the following steps:
the invention provides an SAR image edge feature extraction method based on a combined filter, which comprises the following steps:
performing fusion processing on the index parts of the Gaussian gamma-shaped double-window filter operator and the Gabor filter operator to combine the Gaussian gamma-shaped double-window filter operator and the Gabor filter operator, so as to construct a combined filter operator;
filtering the image by using the constructed combined filter operator to respectively obtain a first filtering value and a second filtering value; calculating the ratio of the first filtering value to the second filtering value and the minimum value of the ratio of the second filtering value to the first filtering value; carrying out Gabor filtering on the image in the filtering window of the constructed combined filtering operator to obtain a third filtering value; and solving the minimum value of the difference value between the third filtering value and the minimum value to obtain the edge characteristics of the image.
The beneficial effects of the above technical scheme are: according to the invention, the Gabor filtering operator and the Gauss gamma-shaped double-window filtering operator are combined to construct a combined filter operator according to the good multi-scale characteristic and multi-directional characteristic of the Gabor filtering operator, and the constructed combined filter operator is utilized to filter the image to obtain the edge characteristic of the image so as to obtain the image after edge extraction. The image can eliminate false response generated by factors such as noise and the like, can also improve the processing effect on a high reflection region in the SAR image, and has a good edge detection effect. Meanwhile, due to the adoption of the mode of combining the Gaussian gamma-shaped double-window filtering operator and the Gabor filtering operator, the influence of SAR on redundant information after filtering can be effectively eliminated.
As a further improvement of the method, if the image is a two-dimensional discrete sequence, the combination filter operator is:
Figure BDA0002371967960000021
Figure BDA0002371967960000022
in the formula (I), the compound is shown in the specification,
Figure BDA0002371967960000023
respectively representing the combined filter operators in the U, L direction; theta is the rotation angle of the counter-clockwise rotation of the operator and is taken as
Figure BDA0002371967960000024
xr＝xcosθ-ysinθ，yrX sin θ + ycos θ; is a gamma function operator; alpha and beta are width control parameters of a filter window of a combined filter operator, sigma is the function bandwidth of a Gaussian gamma-shaped double-window filter operator, and is a length control parameter of the filter window of the combined filter operator; (u)0,v0) And 2P is a preset convolution numerical value, namely the wavelet center frequency in the Gabor filter operator.
As a further improvement of the method, the edge features of the image are:
Figure BDA0002371967960000025
Figure BDA0002371967960000026
Figure BDA0002371967960000031
Figure BDA0002371967960000032
in the formula, xi is an edge feature obtained after a combined filter operator; i is image data of an image; represents a convolution operator; min represents the minimum function operator; m1For use in a direction perpendicular to the rotation angle theta
Figure BDA0002371967960000033
The operator convolutes the image in the filtering window to obtain a first filtering value; m1(x,y|θk) Is the k-th rotation angle thetakA first intermediate parameter of a first filtering value of the corresponding image; m2For use in the horizontal direction at a rotation angle theta
Figure BDA0002371967960000034
The operator convolutes the image in the filtering window to obtain a second filtering value; m2(x,y|θk) Is the k-th rotation angle thetakA second intermediate parameter of a second filtering value of the corresponding image; xikAt a rotation angle thetakA minimum value of a ratio of filtering results between the first filtered value in the vertical direction and the second filtered value in the horizontal direction; psikAt a rotation angle thetakCarrying out Gabor filtering in the direction to obtain a third filtering value; k is 0,1, …, 2P-1.
As a further improvement of the method, the length control parameter of the filter window of the combination filter operator and the length of the filter window of the combination filter operator have a relationship of:
Figure BDA0002371967960000035
where l is the length of the filter window of the combiner filter operator.
As a further improvement of the method, the width control parameter of the combined filter operator filter window and the width of the combined filter operator filter window have a relationship of:
Figure BDA0002371967960000036
where w is the width of the filter window of the combiner filter operator.
As a further improvement of the method, in order to solve the problem that only a plurality of discontinuous short straight line segments can be extracted and false responses exist, the method further comprises the step of performing edge optimization on the obtained edge features.
As a further improvement of the method, in order to extract and express the edge structure of the image more completely, the edge optimization includes the following steps: traversing the extracted straight line segments, and merging two straight line segments of which the distance between the straight line segments is smaller than a set distance threshold to obtain merged straight line segments; the distance between the straight line segments comprises a space distance between two straight lines and a straight line distance, and the corresponding set distance threshold values are a set first space distance threshold value and a set straight line distance threshold value respectively; the spatial distance is the maximum value of the distances from two endpoints of the shorter straight line segment to the longer straight line segment; the distance between the two straight lines is the maximum value of the distance from the two end points of one straight line section to the two end points of the other straight line section.
As a further improvement of the method, the merging process comprises: deleting the shorter straight line segment; processing is carried out according to the positions of two end points of the shorter straight line to two vertical feet of the longer straight line section and the positions of two end points of the longer straight line section: if the two vertical feet are positioned on the long straight line segment, taking the longer straight line segment as a new straight line segment; and if the position of the vertical foot is outside the long straight line segment, taking the vertical foot outside the long straight line segment as the endpoint of a new straight line segment.
As a further improvement of the method, in order to perform more complete extraction and expression on the edge structure of the image, after obtaining the combined straight line segment, the method further includes: traversing the merged straight line segments, and prolonging the two straight line segments of which the spatial distance between the straight line segments is less than a set second spatial threshold and the slope difference of the straight line segments is less than a set slope threshold.
As a further improvement of the method, the extension treatment comprises: connecting the nearest ends of the two straight line segments.
The invention also provides an edge feature extraction device of the SAR image based on the combined filter, which comprises a memory and a processor, wherein the processor is used for executing instructions stored in the memory to realize the above introduced edge feature extraction method of the SAR image based on the combined filter, and can achieve the same effect as the method.
Drawings
FIG. 1 is a flow chart of a method of the present invention;
FIG. 2 is a flow chart of the edge optimization process (i.e., the modified LSD algorithm) of the present invention;
FIG. 3 is a schematic illustration of two straight line segments ab, cd of the present invention;
FIG. 4-1 is a schematic diagram of the results of the linear extraction of the original SAR image by the LSD algorithm in the prior art;
FIG. 4-2 is a schematic diagram of the results of processing an original SAR image using a prior art Gabor filter operator;
FIG. 4-3 is a schematic diagram of the results of processing an original SAR image using the combined filter operator of the present invention;
4-4 are schematic diagrams of results of edge optimization processing using a prior art LSD algorithm with respect to the results shown in FIGS. 4-3;
fig. 4-5 are schematic diagrams of the results of the edge optimization process using the improved LSD algorithm of the present invention with respect to the results shown in fig. 4-3.
Detailed Description
The method comprises the following steps:
the embodiment provides an edge feature extraction method of an SAR image based on a combined filter, which is applied to edge feature detection of the SAR image. The method is described in detail below with reference to fig. 1.
Step one, performing fusion processing on index parts of a Gaussian gamma-shaped double-window filter operator and a Gabor filter operator to combine the Gaussian gamma-shaped double-window filter operator and the Gabor filter operator, and thus constructing and obtaining a combined filter operator.
The Gaussian gamma-shaped double-window filter operator is an edge detection operator based on a ratio, is mainly used for extracting an intensity map of an image edge, and can effectively remove false responses near the image edge. The Gabor filtering operator can fully cover the image in a frequency domain, has good multi-scale characteristics and multidirectional characteristics, and can extract effective characteristics in different scales and directions from a target image. Therefore, the combined filter operator is constructed by combining the two characteristics. It should be noted that the computation complexity of the operator of the combined filter is large, and in the process of dealing with large image processing, a multithreading block parallel processing method can be used in combination to shorten the filtering time and improve the processing efficiency. The constructed combined filter operator is:
Figure BDA0002371967960000051
Figure BDA0002371967960000052
in the formula (I), the compound is shown in the specification,
Figure BDA0002371967960000053
respectively representing the combined filter operators in the U, L direction; theta is the rotation angle of the combined filter operator in counterclockwise rotation and is taken as
Figure BDA0002371967960000054
xr＝xcosθ-ysinθ，yrX sin θ + ycos θ; is a gamma function operator; alpha and beta are width control parameters of a filter window of the combined filter operator, and sigma is a length control parameter of the filter window of the combined filter operator; (u)0,v0) The wavelet center frequency in the Gabor operator.
And step two, filtering the image by using the constructed combined filter operator, and obtaining the edge characteristic xi of the image according to the steps from the formula (3) to the formula (6). Specifically, the method comprises the following steps: for each rotation angle theta, performing 2P-time convolution on the SAR image in the filtering window corresponding to the combined filter operator by using the formula (1) and the formula (2) to obtain:
Figure BDA0002371967960000055
Figure BDA0002371967960000056
Figure BDA0002371967960000057
ξ＝mink＝0,1,…,2P-1(ψk-ξk) (6)
in the formula, xi is an edge feature obtained after a combined filter operator; i is image data of the SAR image; represents a convolution operator; min represents the minimum function operator; m1For use in a direction perpendicular to the rotation angle theta
Figure BDA0002371967960000058
The operator convolves the SAR image in the filtering window to obtain a first filtering value; m1(x,y|θk) Is the k-th rotation angle thetakA first intermediate parameter of a first filtering value of the corresponding SAR image; m2For use in the horizontal direction at a rotation angle theta
Figure BDA0002371967960000061
The operator convolves the SAR image in the filtering window to obtain a second filtering value; m2(x,y|θk) Is the k-th rotation angle thetakA second intermediate parameter of a second filtering value of the corresponding SAR image; sigma is the function bandwidth of a Gaussian gamma-shaped double-window filter operator and is used for controlling the length l of a filter window of the combined filter operator; xikAt a rotation angle thetakA minimum value of a ratio of filtering results between the first filtered value in the vertical direction and the second filtered value in the horizontal direction; psikAt a rotation angle thetakCarrying out Gabor filtering on the image in the filtering window in the direction to obtain a third filtering value; k is 0,1, …,2P-1, 2P is a preset volumeThe integral number.
The relationship between the length l of the filter window and the length control parameter σ, and the relationship between the width w of the filter window and the width control parameters α and β are:
Figure BDA0002371967960000062
Figure BDA0002371967960000063
wherein alpha is a preset fixed integer value and is more than or equal to 2.
And step three, performing edge optimization on the obtained edge features by adopting an improved LSD algorithm, wherein a specific flow chart of the improved LSD algorithm is shown in FIG. 2. The method specifically comprises the following steps:
step 1, merging treatment: traversing a first straight line segment set obtained after filtering by a combined filter operator, judging whether the spatial distance between two straight line segments is smaller than a set first spatial distance threshold (which can be set to 20 pixels) and the distance between the two straight line segments is smaller than a set straight line distance threshold (which can be set to 30 pixels), if so, combining the two straight line segments to obtain combined straight line segments, continuously repeating the step 1 until the situation does not exist in the first straight line segment set, and finally forming a second straight line segment set.
Wherein the spatial distance between two straight line segments defined herein is: the maximum of the distances of the two endpoints of the shorter of the two straight line segments from the longer of the two straight line segments. For example, as shown in fig. 3, two straight line segments are ab and cd, respectively, and ab is the longer of the two straight line segments. The spatial distance between these two straight line segments is then: the maximum of the distance from the end point c to the straight line segment ab and the distance from the end point d to the straight line segment ab.
The straight line distance between two straight line segments defined herein is: the maximum value of the distance between the two end points of one straight line segment and the two end points of the other straight line segment. For example, as shown in FIG. 3, the straight lines of the straight line segments ab, cd are spaced by: ac. The maximum of ad, bc, bd.
When two straight line segments are identified as the same type, the specific merging means is as follows: taking a vertical line from the shorter straight line segment to the longer straight line segment to obtain two vertical feet, and comparing the positions of the two vertical feet with the positions of two end points of the longer straight line segment: if the two drop foot positions are both between two end points of the longer straight-line segment, the original longer straight-line segment is kept as a new straight-line segment, and the original shorter straight-line segment is deleted; if the positions of the drop feet are outside the two end points of the longer straight-line segment, taking the positions of the drop feet outside the two end points of the longer straight-line segment as the end points of a new straight-line segment, thereby forming a new straight-line segment, namely, lengthening the longer straight-line segment according to the positions of the two drop feet and the positions of the two end points of the longer straight-line segment. Namely: and taking the point with the minimum X coordinate in the four points as the starting point of a new straight line segment, taking the point with the maximum X coordinate as the end point of the new straight line segment, and deleting the shorter straight line segment.
Step 2, prolonging treatment: and traversing the second straight-line segment set, judging whether the spatial distance between two straight-line segments is smaller than a set second spatial distance threshold (which can be set to 80 pixels) and the slope difference of the two straight-line segments is smaller than a set slope threshold (which can be set to 0.01), and if so, merging the two straight-line segments. Step 2 is repeated until there is no such event in the second set of straight line segments.
The specific extension means is as follows: the nearest end points of the two straight line segments are connected to obtain a new straight line segment.
In this embodiment, step three is to perform edge optimization on the edge feature. In order to realize edge optimization, the edge optimization may be performed according to the existing LSD algorithm instead of the third step in the above embodiment, but the effect is not as good as the improved LSD algorithm proposed in this embodiment. Moreover, the specific edge optimization includes a merging process and an extension process (corresponding to step 1 and step 2, respectively). In another embodiment, the merging process may be performed only on the straight line segments, and the extension process may not be performed. Of course, the specific combination and extension means set forth in this embodiment may not be employed. For example, the specific merging means and the extension means are both to fit two straight line segments into one straight line segment.
Moreover, the method can be applied to other images with similar problems.
Different methods are used to process a specific SAR image.
Fig. 4-1 is a schematic diagram of a result obtained by performing linear extraction on an original SAR image by using the LSD algorithm in the prior art. As can be seen from the figure, only by using the LSD algorithm of the prior art, only a few short straight lines which are difficult to form structural features can be extracted, and cannot be effectively utilized in the next processing. Therefore, the SAR image needs to be filtered before this processing.
Fig. 4-2 is a schematic diagram of the result of processing the original SAR image by using the Gabor filter operator in the prior art. It can be seen from the figure that due to the non-orthogonality of the Gabor filter operator, the processed image generates much redundant interference information, so that the background and the edge structural features are difficult to distinguish, and the structural features still stay at the visual level and are difficult to extract.
Fig. 4-3 is a schematic diagram of the results of processing the original SAR image using the combined filter operator of the present invention. It can be seen from the figure that redundant information can be greatly reduced by adopting a combined filter operator, the edge structure characteristics of the original SAR image and the integral image brightness information are well maintained, and preparation work is made for extracting a continuous edge structure in the next step.
Fig. 4-4 is a diagram illustrating the results of line extraction using the prior art LSD algorithm with respect to the results of fig. 4-3. It can be seen from the figure that although the filtering result can present more straight line features, the feature extraction is performed by using the existing LSD algorithm, only more discontinuous short straight line segments are extracted, only approximate expression can be formed on the structural features, and some false responses exist under the influence of noise factors.
Fig. 4-5 are schematic diagrams of the results of the straight line extraction using the improved LSD algorithm of the present invention with respect to the results of fig. 4-3. It can be seen from the figure that after the improved LSD algorithm is used for edge optimization, long and continuous straight line features are formed, the edge structure of the image can be extracted and expressed more completely, and a real and reliable basis is provided for the subsequent processing of the image.
The embodiment of the device is as follows:
the embodiment provides an edge feature extraction device of a SAR image based on a combined filter, which comprises a memory and a processor, wherein the device comprises the memory and the processor, and the memory and the processor are directly or indirectly electrically connected to realize data transmission or interaction. The processor may be a general-purpose processor, such as a central processing unit CPU, or may be another programmable logic device, such as a digital signal processor DSP, and the processor is configured to execute instructions stored in a memory to implement the method for extracting edge features of the SAR image based on the combined filter described in the method embodiment. The method has been described in detail in the method embodiments, and is not described herein again.

Claims (9)

1. An SAR image edge feature extraction method based on a combined filter is characterized by comprising the following steps:
performing fusion processing on the index parts of the Gaussian gamma-shaped double-window filter operator and the Gabor filter operator to combine the Gaussian gamma-shaped double-window filter operator and the Gabor filter operator, so as to construct a combined filter operator;
filtering the image by using the constructed combined filter operator to respectively obtain a first filtering value and a second filtering value; calculating the ratio of the first filtering value to the second filtering value and the minimum value of the ratio of the second filtering value to the first filtering value; carrying out Gabor filtering on the image in the filtering window of the constructed combined filtering operator to obtain a third filtering value; calculating the minimum value of the difference value between the third filtering value and the minimum value to obtain the edge characteristic of the image;
if the image is a two-dimensional discrete sequence, the combined filter operator is:
Figure FDA0002780931930000011
Figure FDA0002780931930000012
in the formula (I), the compound is shown in the specification,
Figure FDA0002780931930000013
respectively representing the combined filter operators in the U, L direction; theta is the rotation angle of the counter-clockwise rotation of the operator and is taken as
Figure FDA0002780931930000014
xr＝xcosθ-ysinθ，yrX sin θ + ycos θ; is a gamma function operator; alpha and beta are width control parameters of a filter window of a combined filter operator, sigma is the function bandwidth of a Gaussian gamma-shaped double-window filter operator, and is a length control parameter of the filter window of the combined filter operator; (u)0,v0) The central frequency of the wavelet in the Gabor filter operator is shown, and 2P is a preset convolution numerical value;
Figure FDA0002780931930000015
wherein I is image data of the image, representing a convolution operator, M1For use in a direction perpendicular to the rotation angle theta
Figure FDA0002780931930000016
The operator convolves the image in the filter window to obtain a first filter value M2For use in the horizontal direction at a rotation angle theta
Figure FDA0002780931930000017
And the operator convolutes the image in the filtering window to obtain a second filtering value.
2. The method of claim 1, wherein the edge features of the image are as follows:
Figure FDA0002780931930000018
Figure FDA0002780931930000019
Figure FDA0002780931930000021
in the formula, xi is an edge feature obtained after a combined filter operator; min represents the minimum function operator; m1(x,y|θk) Is the k-th rotation angle thetakA first intermediate parameter of a first filtering value of the corresponding image; m2(x,y|θk) Is the k-th rotation angle thetakA second intermediate parameter of a second filtering value of the corresponding image; xikAt a rotation angle thetakA minimum value of a ratio of filtering results between the first filtered value in the vertical direction and the second filtered value in the horizontal direction; psikAt a rotation angle thetakCarrying out Gabor filtering in the direction to obtain a third filtering value; k is 0,1, …, 2P-1.
3. The method for extracting edge features of an SAR image based on a combined filter according to claim 1 or 2, wherein the relationship between the length control parameter of the filtering window of the combined filter operator and the length of the filtering window of the combined filter operator is:
Figure FDA0002780931930000022
where l is the length of the filter window of the combiner filter operator.
4. The method for extracting edge features of an SAR image based on a combined filter according to claim 1 or 2, wherein the relation between the width control parameter of the filtering window of the combined filter operator and the width of the filtering window of the combined filter operator is:
Figure FDA0002780931930000023
where w is the width of the filter window of the combiner filter operator.
5. The method for extracting edge features of an SAR image based on a combined filter according to claim 1, further comprising a step of performing edge optimization on the obtained edge features.
6. The method of claim 5, wherein the edge optimization comprises the following steps: traversing the extracted straight line segments, and merging two straight line segments of which the distance between the straight line segments is smaller than a set distance threshold to obtain merged straight line segments;
the distance between the straight line segments comprises a space distance between two straight lines and a straight line distance, and the corresponding set distance threshold values are a set first space distance threshold value and a set straight line distance threshold value respectively;
the spatial distance is the maximum value of the distances from two endpoints of the shorter straight line segment to the longer straight line segment;
the distance between the two straight lines is the maximum value of the distance from the two end points of one straight line section to the two end points of the other straight line section.
7. The method of claim 6, wherein the merging process comprises: deleting the shorter straight line segment; processing is carried out according to the positions of two end points of the shorter straight line to two vertical feet of the longer straight line section and the positions of two end points of the longer straight line section: if the two vertical feet are positioned on the long straight line segment, taking the longer straight line segment as a new straight line segment; and if the position of the vertical foot is outside the long straight line segment, taking the vertical foot outside the long straight line segment as the endpoint of a new straight line segment.
8. The method for extracting edge features of an SAR image based on a combined filter according to claim 6 or 7, wherein after obtaining the combined straight line segment, the method further comprises: traversing the merged straight line segments, and prolonging the two straight line segments of which the spatial distance between the straight line segments is less than a set second spatial threshold and the slope difference of the straight line segments is less than a set slope threshold.
9. The method of claim 8, wherein the extension process comprises: connecting the nearest ends of the two straight line segments.
CN202010053326.6A 2020-01-17 2020-01-17 SAR image edge feature extraction method based on combined filter Expired - Fee Related CN111402279B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010053326.6A CN111402279B (en) 2020-01-17 2020-01-17 SAR image edge feature extraction method based on combined filter

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010053326.6A CN111402279B (en) 2020-01-17 2020-01-17 SAR image edge feature extraction method based on combined filter

Publications (2)

Publication Number Publication Date
CN111402279A CN111402279A (en) 2020-07-10
CN111402279B true CN111402279B (en) 2021-01-01

Family

ID=71430288

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010053326.6A Expired - Fee Related CN111402279B (en) 2020-01-17 2020-01-17 SAR image edge feature extraction method based on combined filter

Country Status (1)

Country Link
CN (1) CN111402279B (en)

Family Cites Families (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105809173B (en) * 2016-03-09 2019-01-29 中南大学 A kind of image RSTN invariable attribute feature extraction and recognition methods based on bionical object visual transform
CN107886505B (en) * 2017-11-08 2019-06-04 电子科技大学 A kind of synthetic aperture radar airfield detection method based on line primitives polymerization
CN108564557B (en) * 2018-05-31 2020-08-25 京东方科技集团股份有限公司 Image correction method and device
CN109410233A (en) * 2018-09-27 2019-03-01 中国科学院沈阳应用生态研究所 A kind of accurate extracting method of high-definition picture road of edge feature constraint
CN110533679B (en) * 2019-07-29 2021-09-03 西安电子科技大学 SAR image edge detection method based on logarithm transformation and gabor convolution

Non-Patent Citations (4)

* Cited by examiner, † Cited by third party
Title
An Advanced Multiscale Edge Detector Based;Yuming Xiang等;《IEEE GEOSCIENCE AND REMOTE SENSING LETTERS》;20170930;第14卷(第9期);1522-1526 *
Edge Detector of SAR Images Using Gaussian-Gamma-Shaped Bi-Windows;P. Shui等;《IEEE Geoscience and Remote Sensing Letters》;20200930;第9卷(第5期);846-850 *
Efficient Edge Detection Using Simplified Gabor Wavelets;W. Jiang等;《IEEE Transactions on Systems, Man, and Cybernetics, Part B (Cybernetics)》;20090831;第39卷(第4期);1036-1047 *
基于改进双边滤波的非均匀光照图像校正算法;郑娜等;《电子测量技术》;20190930;第42卷(第17期);157-161 *

Also Published As

Publication number Publication date
CN111402279A (en) 2020-07-10

Similar Documents

Publication Publication Date Title
Ziou et al. Edge detection techniques-an overview
Truc et al. Vessel enhancement filter using directional filter bank
CN106709928B (en) fast two-dimensional maximum inter-class variance threshold method for noisy images
CN109919870B (en) SAR image speckle suppression method based on BM3D
CN111080661A (en) Image-based line detection method and device and electronic equipment
Li et al. Road lane detection with gabor filters
CN102222322A (en) Multiscale non-local mean-based method for inhibiting infrared image backgrounds
Chai et al. Stereo matching algorithm based on joint matching cost and adaptive window
Krishnan et al. A survey on different edge detection techniques for image segmentation
CN110660065B (en) Infrared fault detection and identification algorithm
CN109146803B (en) SAR image radiation resolution improving method and device based on multi-angle image
CN113129323A (en) Remote sensing ridge boundary detection method and system based on artificial intelligence, computer equipment and storage medium
CN109087347B (en) Image processing method and device
CN111402279B (en) SAR image edge feature extraction method based on combined filter
CN109636822A (en) A kind of improvement Canny adaptive edge extracting method based on new building subordinating degree function
Kumar et al. Image segmentation using discontinuity-based approach
Wu et al. Research on crack detection algorithm of asphalt pavement
Li et al. A study of crack detection algorithm
CN109377507B (en) Hyperspectral remote sensing image segmentation method based on spectral curve spectral distance
Jiao et al. Infrared dim small target detection method based on background prediction and high-order statistics
CN115994870B (en) Image processing method for enhancing denoising
Wang et al. Saliency-based adaptive object extraction for color underwater images
Ji et al. A novel method of image segmentation using watershed transformation
CN115797773A (en) Ultra-wideband radar target identification method based on edge detection and region segmentation
CN112927169B (en) Remote sensing image denoising method based on wavelet transformation and improved weighted kernel norm minimization

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
CF01 Termination of patent right due to non-payment of annual fee

Granted publication date: 20210101

Termination date: 20220117

CF01 Termination of patent right due to non-payment of annual fee