CN106557740A - The recognition methods of oil depot target in a kind of remote sensing images - Google Patents

The recognition methods of oil depot target in a kind of remote sensing images Download PDF

Info

Publication number
CN106557740A
CN106557740A CN201610910833.0A CN201610910833A CN106557740A CN 106557740 A CN106557740 A CN 106557740A CN 201610910833 A CN201610910833 A CN 201610910833A CN 106557740 A CN106557740 A CN 106557740A
Authority
CN
China
Prior art keywords
target
region
image
scale
interest
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.)
Granted
Application number
CN201610910833.0A
Other languages
Chinese (zh)
Other versions
CN106557740B (en
Inventor
孙向东
朱军
杨卫东
赵革
邹腊梅
翟展
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.)
Huazhong University of Science and Technology
Original Assignee
Huazhong University of Science and Technology
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 Huazhong University of Science and Technology filed Critical Huazhong University of Science and Technology
Priority to CN201610910833.0A priority Critical patent/CN106557740B/en
Publication of CN106557740A publication Critical patent/CN106557740A/en
Application granted granted Critical
Publication of CN106557740B publication Critical patent/CN106557740B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00Scenes; Scene-specific elements
    • G06V20/10Terrestrial scenes
    • G06V20/176Urban or other man-made structures

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Multimedia (AREA)
  • Theoretical Computer Science (AREA)
  • Image Analysis (AREA)

Abstract

The invention discloses in a kind of remote sensing images oil depot target recognition methods, calculate the phase spectrum conspicuousness of whole scene first, according to phase spectrum conspicuousness extract scene in the be possible to area-of-interest comprising target;In feature extraction, using the partial structurtes feature of the calculating area-of-interest of local regression nuclear model pointwise, and the Feature Descriptor that can describe object construction is generated;In the target detection stage, similarity measurement is made with cosine similarity, calculate the similarity of area-of-interest and oil depot sample image, and the characteristic in the positive and negative sample separating capacity using Feature Descriptor and similitude face builds the decision networks with adaptive ability, the PRELIMINARY RESULTS of target detection is obtained by the decision networks, unnecessary PRELIMINARY RESULTS is removed by non-maxima suppression algorithm, final target detection result is obtained;In this general remote sensing images proposed by the present invention, oil depot mesh object detection method is good for the target identification effect of multiple dimensioned, various visual angles.

Description

Method for identifying oil reservoir target in remote sensing image
Technical Field
The invention belongs to the technical field of remote sensing image target identification, and particularly relates to a method for identifying an oil reservoir target in a remote sensing image.
Background
The high-resolution remote sensing image provides rich detail information, so that the identification of various specific targets becomes possible; however, factors such as noise interference, seasonal weather, shadow, illumination intensity, and occlusion may cause the structure and texture information of the internal details of the target to fluctuate, which makes the recognition of high-resolution images difficult.
The method for detecting the remote sensing image oil depot target in the prior art comprises the following steps: a target detection method based on deep learning, a target identification detection method based on prior knowledge, and a target detection method based on a model; the target detection method based on deep learning has quite high dependence on sample information richness and sample quantity, and only a simple data image source can be provided for identification of the remote sensing image oil depot target in most cases; the target identification method based on the priori knowledge is to judge the position of a target by using the priori knowledge of the target, such as the average value, the variance, the moment of invariance and other priori characteristics of an airplane, the characteristics of the target need to be accurately expressed, a decision-making method with self-adaptive capacity is needed, and the target detection accuracy is low under the condition that the expression of the priori knowledge is not accurate enough or the decision-making method is not perfect enough; the model-based method comprises the steps of extracting target features through a large number of experiments, marking model parameters of a target to generate hypothesis and predicting target characteristics, measuring a background or a model of the target in actual application, matching the model with the predicted characteristics to achieve certain similarity, and considering the model as the target; the target detection method based on the model has high requirements on modeling accuracy and fault tolerance rate, and has great difficulty in identifying the target of the oil depot under the complex scene of the remote sensing image.
Disclosure of Invention
Aiming at the defects or improvement requirements of the prior art, the invention provides a method for identifying an oil reservoir target in a remote sensing image, which aims to overcome the dependence of the identification technology in the complex scene of the prior remote sensing image on the number of samples, prior characteristics, modeling accuracy and fault tolerance rate and provide the method for identifying the oil reservoir target in the complex scene of the remote sensing image.
To achieve the above object, according to one aspect of the present invention, there is provided a method for identifying an oil reservoir target in a remote sensing image, comprising the steps of:
(1) extracting an interested region by carrying out phase spectrum significance characteristic calculation and significance characteristic region segmentation on the remote sensing image;
(2) obtaining a multi-scale interested area image by sampling the interested area; extracting the characteristics of the template of the oil depot target and the multi-scale interesting region image to obtain a multi-scale characteristic descriptor;
(3) carrying out target search on the multi-scale feature descriptors to obtain multi-scale similar surfaces, and fusing the multi-scale similar surfaces to generate similar surfaces of the region of interest;
constructing a decision network with self-adaptive capacity according to the positive and negative sample measurements of the similarity surface and the feature descriptor of the region of interest;
(4) preliminarily detecting an oil depot target in the remote sensing image according to the decision network; identifying an oil depot target from the primary detection result by adopting a non-maximum suppression algorithm;
the non-maximum suppression algorithm is an algorithm that sorts similarity values of target regions detected from an interested region from large to small, selects the largest target region, excludes the region with the overlapping area exceeding the threshold value, and iterates until the overlapping areas of all the target regions are smaller than the threshold value.
Preferably, the method for identifying the oil reservoir target in the remote sensing image comprises the following substeps in step (1):
(1.1) carrying out scale normalization on a given remote sensing image, converting the normalized image into a frequency domain through Fourier change, and acquiring the phase spectrum characteristic of the frequency domain image; performing inverse Fourier transform on the phase spectrum characteristic to obtain a significance characteristic diagram;
and (1.2) segmenting the significant feature map by adopting a maximum stable region feature extraction method to obtain a plurality of interested regions.
Preferably, the method for identifying the oil reservoir target in the remote sensing image comprises the following substeps in step (2):
(2.1) sampling each interested region upwards and downwards to obtain a multi-scale interested region image;
and (2.2) extracting the characteristics of the template of the oil depot target and the multi-scale interested region image to obtain a template characteristic descriptor and a characteristic descriptor of the multi-scale interested region image.
Preferably, the method for identifying the oil reservoir target in the remote sensing image comprises the following substeps in step (2.2):
(2.2.1) calculating the local regression kernel characteristic W of the interested region image under each scale pixel by pixelj
Wherein:
k (-) is a gaussian function, x refers to the center pixel coordinate of the region of interest image,refers to the two-dimensional space coordinates in the image of the interested area; matrix arrayh represents the smoothing coefficient, matrix ClIs a covariance matrix of the matrix composed of the gradients of each pixel in the integration window at the coordinate p × p;
n is the number of the fragments, the corresponding region of interest is divided into n p multiplied by p fragments, j is the number of the fragments, and p is the size of the integration window;
(2.2.2) connecting the local regression kernel features of each pixel in series to obtain an initial structural feature descriptor of the interested region imageWherein, I refers to an identity matrix, and R refers to a real number;
(2.2.3) extracting the characteristics of the template of the oil depot target and the multi-scale interested region image to obtain a template characteristic descriptor WQAnd region of interest image feature descriptor WT
Wherein,wherein n isTThe number of the target image fragment is referred to;
(2.2.4) template feature descriptor W for the aboveQPerforming dimensionality reduction treatment to obtain the structural characteristics of the oil depot template after dimensionality reduction
Wherein, PCA dimension reduction matrixWherein d is a characteristic dimension after dimension reduction;
(2.2.5) by PCA dimension reduction matrix AQThe image characteristics of the interested region under each scale are describedTThe main component is extracted, and the main component is extracted,
feature descriptors for obtaining images of regions of interest at various scales
Preferably, the method for identifying the oil reservoir target in the remote sensing image comprises the following substeps in step (3):
(3.1) searching a target on the multi-scale feature by using the size of the target template of the oil depot and a fixed step length to obtain a corresponding multi-scale similar surface;
mapping the similar surfaces of all scales to the original scale to obtain the similar surfaces of the original scale, and fusing the similar surfaces of the original scale to obtain similar surfaces of the region of interest;
(3.2) intercepting a plurality of positive and negative samples from a given remote sensing image, and obtaining feature descriptors of the positive and negative samples;
(3.3) carrying out similarity measurement calculation on the feature descriptors of the positive and negative samples and the feature surface of the oil depot template to obtain the positive and negative sample measurements of the feature descriptors; and combining the positive and negative sample metrics with the similarity histogram statistical characteristics of the similar surface of the region of interest to construct a decision network with self-adaptive capacity.
Preferably, the method for identifying the oil reservoir target in the remote sensing image comprises the following substeps in step (3.3):
(3.3.1) calculating local regression kernel characteristics of each positive sample and each negative sample of the remote sensing image, and performing correlation calculation with the characteristic surface of the oil depot template to obtain frequency normalization curves of correlation coefficients of the positive samples and the negative samples; taking the abscissa of the intersection point of the frequency normalization curve of positive samples and the frequency normalization curve of negative samples as a first decision threshold τ0
(3.3.2) for fusionThe similarity surface of the interested region takes the correlation coefficient with the relation number Kth maximum in the similarity surface as a threshold value tau to be selected, and takes the tau1=max(τ,τ0) As a second decision threshold;
(3.3.3) determining from the first decision threshold τ0And a second decision threshold τ1Forming the decision network.
Preferably, the method for identifying the oil reservoir target in the remote sensing image comprises the following substeps in step (4):
(4.1) carrying out similarity surface screening through the decision network, and when the correlation coefficient in the similarity surface is larger than the first threshold value tau0If so, judging that the target exists in the similar surface; when the correlation coefficient of the interested area is larger than the second threshold value tau1Judging the region of interest as a target region;
(4.2) identifying reservoir targets from the target area using a non-maxima suppression algorithm.
Preferably, the method for identifying the oil reservoir target in the remote sensing image comprises the following substeps in step (4.2):
(4.2.1) sequencing all the target areas from high to low according to the similarity values, and determining the target area with the maximum similarity value;
(4.2.2) acquiring the overlapping area of the target area with the maximum similarity value and all the target areas;
(4.2.3) removing the target area with the overlapping area larger than the area threshold;
(4.2.4) repeating the steps (4.2.1) - (4.2.3) until the overlapping areas of all the target areas are smaller than a preset area threshold; and the ratio of the intersection and the union of the two overlapped target regions is smaller than 0.3 through a preset area threshold value.
In general, compared with the prior art, the above technical solution contemplated by the present invention can achieve the following beneficial effects:
(1) according to the method for identifying the oil depot target in the remote sensing image, the structural characteristics of the oil depot target are considered in the processing process of the step (3), a local control kernel is used as the characteristic description of the image, and the characteristic can well distinguish the target from the background because the local control kernel characteristic of the target of the oil depot has good positive and negative sample distinguishing capacity;
(2) according to the method for identifying the oil depot target in the remote sensing image, the remote sensing image is preprocessed by extracting the region of interest, so that the search range of the oil depot target is reduced, the interference of a complex background on information is reduced, the calculated amount is greatly reduced, and the processing instantaneity is improved;
(3) according to the method for identifying the oil reservoir target in the remote sensing image, the region of interest can extract the structured feature descriptors which are accurate and have good sample distinguishing capability, so that the technical problems of low detection accuracy and high false alarm rate caused by multiple types of targets in the same scene with complex scene, multiple viewpoints and multiple scales in the process of detecting the oil reservoir target are solved.
Drawings
FIG. 1 is a schematic flow chart of a method for identifying an oil reservoir target in a remote sensing image according to an embodiment of the present invention;
fig. 2 is a schematic diagram illustrating comparison of results of a base station similarity plane calculation method according to an embodiment of the present invention; wherein (a) is to obtain a feature matrix f by the similarity value rhoQAnd fTThe relevant face of (a); (b) the similar surfaces are obtained by fusing the regions of interest in the embodiment;
FIG. 3 is a schematic diagram of a result of remote sensing image oil depot target identification according to an embodiment of the invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention. In addition, the technical features involved in the embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other.
The method for identifying the oil reservoir target in the remote sensing image comprises the steps of extracting an interested area in the remote sensing image, calculating phase spectrum characteristics, extracting a salient area according to the phase spectrum characteristics, and obtaining information such as coordinates, size and the like of the interested area; then according to the obtained region of interest, carrying out feature description on the region of interest based on a local regression kernel; finally, respectively extracting the characteristics of the oil depot sample image and the target interesting area, calculating the similarity between all target areas and the oil depot sample image, constructing a decision network with self-adaptive capacity, and finally performing non-maximum suppression on the result of the decision network to obtain a target detection result;
the specific flow is as illustrated in fig. 1, and comprises the following steps:
(1) carrying out scale normalization on the remote sensing image, and then carrying out phase spectrum significance characteristic calculation and row significance region extraction;
segmenting the salient feature region by using a maximum stable region feature extraction method to obtain a plurality of small interested region images;
(2) sampling the interested region image upwards and downwards to obtain interested region images under multiple scales;
(3) calculating Local regression kernel (LSK) characteristics of the interested region image under each scale pixel by pixel to obtain characteristic surfaces of the interested region image under each scale to obtain a series of multi-scale characteristic surfaces;
(4) performing target search on the series of multi-scale feature surfaces according to the fixed target frame size and the fixed step length to obtain corresponding multi-scale similar surfaces;
mapping the similar surfaces of all scales to the original scale, and obtaining similar surfaces of the region of interest through fusion;
(5) intercepting a plurality of positive and negative samples from the remote sensing image, and calculating feature descriptors (LSKs) of the positive and negative samples; the size of the sample is consistent with that of the target template of the oil depot;
similarity measurement calculation is carried out on the positive and negative samples and a target model of the oil depot to obtain positive and negative sample measurements of the feature descriptors; constructing a decision network with self-adaptive capacity by combining the similarity histogram statistical characteristics of the similar surfaces of the interested region;
(6) performing primary detection on the oil depot targets in the scene according to a decision network to obtain a primary target detection result, wherein the primary target detection result comprises whether the oil depot targets exist or not and the number of the oil depot targets;
(7) and identifying the target of the oil depot according to the preliminary detection result by using a non-maximum suppression algorithm.
The method for identifying the oil reservoir target in the remote sensing image provided by the embodiment specifically comprises the following steps:
(1) preprocessing an image and extracting an interested area; the method comprises the following substeps:
(1.1) for a given remote sensing image, firstly carrying out scale normalization, and then carrying out significance characteristic calculation of a phase spectrum to obtain a significance characteristic diagram;
(1.2) on the basis of the remote sensing image saliency characteristic map, segmenting the saliency characteristic map into small interesting regions by adopting a maximum stable region characteristic extraction method;
the maximum stable region feature extraction method is to perform binarization processing on a gray level image by adopting a plurality of thresholds; in all the obtained binary images, some connected regions in the images have little or no change, and the region is the obtained maximum stable extremum region and is used as an interested region; wherein, the threshold value is sequentially increased from 0 to 255 according to the gray value of the image;
(2) extracting LSK multi-scale features;
in the step, the scale transformation is carried out on the region of interest so as to meet the multi-scale adaptability of the target detection of the oil depot; firstly, the method is carried out in a scale range, and an original interested area image I is obtained0Sampling upwards and downwards to obtain Ik=λI0λ is a factor of scale change;
then, extracting the characteristics of the template of the oil depot target and the multi-scale interesting region image to obtain a characteristic descriptor; the method specifically comprises the following substeps:
(2.1) structural feature descriptor calculation:
(2.1.1) calculating the LSK feature W of each fragment one by onej
Wherein:
wherein: k (-) is a gaussian function, x denotes the center pixel coordinate,representing two-dimensional spatial coordinates in the image; matrix arrayh represents the smoothing coefficient, matrix ClIs a covariance matrix of the matrix composed of the gradients of each pixel in the integration window at the coordinate p × p;
(2.1.2) obtaining the LSK characteristic series of each fragment, obtaining an initial structural characteristic descriptor LSKs of the image:
(2.1.3) extracting the characteristics of the template of the oil depot target and the multi-scale interested region image according to the structural characteristic descriptor to obtain a primary template characteristic descriptor WQAnd a preliminary region-of-interest image descriptor WT(ii) a Both are represented by the column vector as follows:
(2.2) analyzing the main components of the structural features;
(2.2.1) template feature descriptor WQPerforming dimension reduction processing, and reserving the characteristic vector of the previous d dimension to obtain a PCA matrixAnd structural characteristics of the oil depot template after dimension reduction
Dimension reduction matrix A by PCAQFor W at various scalesTThe main component is extracted, and the main component is extracted,
obtaining images of various scalesFeature descriptors
In the step, the PCA dimension reduction method is adopted to reserve more obvious and essential characteristics and reduce the complexity of calculation.
(3) Detecting an oil depot target; the method specifically comprises the following substeps:
(3.1) calculating a similarity surface;
by the formulaAnd performing similarity measurement, wherein rho is cosine similarity, and a calculation formula of a high-dimensional structural feature cosine similarity matrix is as follows:
wherein
Computing all fragments in an imageObtaining:
whereinAndis the l-th feature vectorThe jth element of (1); as shown in FIG. 2(a), the feature matrix f is obtained from the similarity values ρQAnd fTThe relevant plane of (2).
(3.2) multi-scale target searching, comprising the following sub-steps:
(3.2.1) structural feature F of oil depot templateQFeature descriptors F of images at various scalesTSearching the oil depot target, and calculating to obtain similar surface RM under each scalek
(3.2.2) mapping each similar surface to the original scale to obtain the similar surface of the original scale, and fusing the similar surfaces of the original scale to obtain the final similar surface RM0
As shown in fig. 2(b), it is a similar surface obtained by the region of interest fusion in the embodiment, as shown in the figure, there are possible three reservoir targets;
(3.3) constructing an adaptive decision network; the key to constructing an adaptive decision network is to determine the first threshold τ0And a second threshold τ1(ii) a First threshold τ0For determining whether a target is present in a scene: when the correlation coefficient is larger than tau in the similarity plane0If yes, the target is considered to exist; second threshold τ1For determining how many targets are present in a scene: when the correlation coefficient of the interested area is more than tau1If so, the region of interest is considered to be an oil depot target region;
the method comprises the following substeps:
(3.3.1) respectively calculating the LSKs of each positive sample and each negative sample, carrying out correlation analysis on the LSKs and the characteristic surface of the oil depot template, carrying out statistics on frequency normalization curves of correlation coefficients of the positive samples and the negative samples, and selecting the abscissa of the intersection point of the positive sample normalization curve and the negative sample normalization curve as a first decision threshold tau0
(3.3.2) statistics of fused similar surface RM0InThe similarity value is calculated by histogram of correlation coefficient of the similarity surface, the correlation coefficient with the Kth maximum correlation number in the similarity surface is taken as a threshold value tau to be selected, and tau is taken as1=max(τ,τ0) As a second decision threshold; in the examples, the coefficient K is 98%;
(3.3.3) passing through the first threshold τ0And a second threshold τ1Carrying out similar surface screening when the correlation coefficient in the similar surface is larger than tau0If so, judging that the target exists in the similar surface; when the correlation coefficient of a point (suspected object) in the region of interest is greater than a first threshold τ1If so, judging that the similar surface is an oil depot target area, wherein a target exists;
(3.4) non-maximum suppression processing:
carrying out non-maximum inhibition treatment on the similar surface with the target screened by the decision network to obtain a target detection result of the oil depot; the method specifically comprises the following substeps:
(3.4.1) sorting all the target areas according to the similarity values from high to low, and determining the target area with the maximum similarity value;
(3.4.2) calculating the overlapping area of the target region with the maximum similarity value and all the target regions;
(3.4.3) removing the target area with the overlapping area larger than the area threshold; in the embodiment, the area threshold is 0.4;
(3.4.4) repeating the steps (3.4.1) to (3.4.3) until the overlapping areas of all the target areas are smaller than the area threshold;
in the embodiment, a non-maximum suppression algorithm is used for screening the primary target detection results obtained through a decision network to obtain oil depot target detection results of a series of scenes as shown in fig. 3, and oil depot targets are marked by white frames; as can be seen from the figure, the detection accuracy rate of the target detection result of the oil depot is high, and the false alarm rate is almost zero.
It will be understood by those skilled in the art that the foregoing is only a preferred embodiment of the present invention, and is not intended to limit the invention, and that any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the scope of the present invention.

Claims (8)

1. A method for identifying an oil reservoir target in a remote sensing image is characterized by comprising the following steps:
(1) extracting an interested region by carrying out phase spectrum significance characteristic calculation and significance characteristic region segmentation on the remote sensing image;
(2) obtaining a multi-scale region-of-interest image by sampling the region-of-interest; extracting the characteristics of the template of the oil depot target and the multi-scale interested region image to obtain a multi-scale characteristic descriptor;
(3) performing target search on the feature descriptors to obtain multi-scale similar surfaces, and fusing the multi-scale similar surfaces to generate similar surfaces of the region of interest;
constructing a decision network with self-adaptive capacity according to the similar surface of the region of interest and the positive and negative sample measures of the feature descriptor;
(4) performing primary detection on an oil depot target in the remote sensing image according to the decision network; and identifying the reservoir target from the result of the primary detection by adopting a non-maximum suppression algorithm.
2. An identification method as claimed in claim 1, characterized in that said step (1) comprises the sub-steps of:
(1.1) carrying out scale normalization on a given remote sensing image, converting the normalized image into a frequency domain through Fourier change, and acquiring the phase spectrum characteristic of the frequency domain image; performing inverse Fourier transform on the phase spectrum characteristic to obtain a significance characteristic diagram;
and (1.2) segmenting the significant feature map by adopting a maximum stable region feature extraction method to obtain a plurality of interested regions.
3. An identification method as claimed in claim 1 or 2, characterized in that said step (2) comprises the sub-steps of:
(2.1) sampling each interested region upwards and downwards to obtain a multi-scale interested region image;
and (2.2) extracting the characteristics of the template of the oil depot target and the multi-scale interested region image to obtain a template characteristic descriptor and a characteristic descriptor of the multi-scale interested region image.
4. An identification method as claimed in claim 3, characterized in that said step (2.2) comprises the sub-steps of:
(2.2.1) calculating the local regression kernel characteristic W of the interested region image under each scale pixel by pixelj
W j ( x l - x ) = K j ( x l - x ; H l ) Σ l = 1 p 2 K j ( x l - x ; H l ) , j = 1 , ... , n l = 1 , ... , p 2 ;
Wherein:
k (-) is a gaussian function, x refers to the center pixel coordinate of the region of interest image,refers to the two-dimensional space coordinates in the image of the interested area; matrix arrayh represents a smoothing coefficient; matrix ClN is the number of the fragments, j is the number of the fragments;
(2.2.2) connecting the local regression kernel features of each pixel in series to obtain an initial structural feature descriptor of the interested region imageWherein, I refers to an identity matrix, and R refers to a real number;
(2.2.3) extracting the characteristics of the template of the oil depot target and the multi-scale interested region image to obtain a template characteristic descriptor WQAnd region of interest image feature descriptor WT
Wherein,wherein n isTThe number of the target image fragment is referred to;
(2.2.4) feature descriptor W for said templateQPerforming dimensionality reduction treatment to obtain the structural characteristics of the oil depot template after dimensionality reduction
Wherein, PCA dimension reduction matrixd is the characteristic dimension after dimension reduction;
(2.2.5) dimensionality reduction of matrix A by the PCAQThe image characteristics of the interested region under each scale are describedTThe main component is extracted, and the main component is extracted,
feature descriptors for obtaining images of regions of interest at various scales
5. An identification method as claimed in claim 1, characterized in that said step (3) comprises the sub-steps of:
(3.1) searching a target on the multi-scale feature by using the size of the target template of the oil depot and a fixed step length to obtain a corresponding multi-scale similar surface; mapping the similar surfaces of all scales to the original scale to obtain the similar surfaces of the original scale, and fusing the similar surfaces of the original scale to obtain similar surfaces of the region of interest;
(3.2) intercepting a plurality of positive and negative samples from a given remote sensing image and obtaining feature descriptors of the positive and negative samples;
(3.3) carrying out similarity measurement calculation on the feature descriptors of the positive and negative samples and the feature surface of the oil depot template to obtain the positive and negative sample measurements of the feature descriptors; and combining the positive and negative sample metrics with the similarity histogram statistical characteristics of the similar surface of the region of interest to construct a decision network with self-adaptive capacity.
6. An identification method as claimed in claim 5, characterized in that said step (3.3) comprises the sub-steps of:
(3.3.1) calculating local regression kernel characteristics of each positive sample and each negative sample of the remote sensing image, and performing correlation calculation with the characteristic surface of the oil depot template to obtain frequency normalization curves of correlation coefficients of the positive samples and the negative samples; taking the abscissa of the intersection point of the frequency normalization curve of positive samples and the frequency normalization curve of negative samples as a first decision threshold τ0
(3.3.2) regarding the similarity surface of the region of interest obtained by fusion, taking the correlation coefficient with the Kth largest correlation number in the similarity surface as a threshold value tau to be selected, and taking tau as the threshold value tau to be selected1=max(τ,τ0) As a second decision threshold;
(3.3.3) determining from the first decision threshold τ0And a second decision threshold τ1Forming the decision network.
7. An identification method as claimed in claim 6, characterized in that said step (4) comprises the sub-steps of:
(4.1) carrying out similarity surface screening through the decision network, and when the correlation coefficient in the similarity surface is larger than the first threshold value tau0If so, judging that the target exists in the similar surface; when the correlation coefficient of the interested area is larger than the second threshold value tau1Judging the region of interest as a target region;
(4.2) identifying reservoir targets from the target area using a non-maxima suppression algorithm.
8. An identification method as claimed in claim 7, characterized in that said step (4.2) comprises the sub-steps of:
(4.2.1) sequencing all the target areas from high to low according to the similarity values, and determining the target area with the maximum similarity value;
(4.2.2) acquiring the overlapping area of the target area with the maximum similarity value and all the target areas;
(4.2.3) removing the target area with the overlapping area larger than the area threshold;
(4.2.4) repeating steps (4.2.1) - (4.2.3) until the overlapping area of all target regions is less than the area threshold.
CN201610910833.0A 2016-10-19 2016-10-19 The recognition methods of oil depot target in a kind of remote sensing images Active CN106557740B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201610910833.0A CN106557740B (en) 2016-10-19 2016-10-19 The recognition methods of oil depot target in a kind of remote sensing images

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201610910833.0A CN106557740B (en) 2016-10-19 2016-10-19 The recognition methods of oil depot target in a kind of remote sensing images

Publications (2)

Publication Number Publication Date
CN106557740A true CN106557740A (en) 2017-04-05
CN106557740B CN106557740B (en) 2019-05-03

Family

ID=58443331

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201610910833.0A Active CN106557740B (en) 2016-10-19 2016-10-19 The recognition methods of oil depot target in a kind of remote sensing images

Country Status (1)

Country Link
CN (1) CN106557740B (en)

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107704509A (en) * 2017-08-31 2018-02-16 北京联合大学 A kind of method for reordering for combining stability region and deep learning
CN108734158A (en) * 2017-04-14 2018-11-02 成都唐源电气股份有限公司 A kind of real-time train number identification method and device
CN108805057A (en) * 2018-05-29 2018-11-13 北京师范大学 A kind of SAR image oil depot area detection method based on joint significance analysis
CN108985288A (en) * 2018-07-17 2018-12-11 电子科技大学 A kind of SAR image oil spilling detection method based on TGMSERs
CN109190457A (en) * 2018-07-19 2019-01-11 北京市遥感信息研究所 A kind of oil depot complex target rapid detection method based on large format remote sensing images
CN110674778A (en) * 2019-09-30 2020-01-10 安徽创世科技股份有限公司 High-resolution video image target detection method and device
CN113837270A (en) * 2021-09-18 2021-12-24 广东人工智能与先进计算研究院 Target identification method, device, equipment and storage medium

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102043958A (en) * 2010-11-26 2011-05-04 华中科技大学 High-definition remote sensing image multi-class target detection and identification method
US20110142282A1 (en) * 2009-12-14 2011-06-16 Indian Institute Of Technology Bombay Visual object tracking with scale and orientation adaptation
CN104463248A (en) * 2014-12-09 2015-03-25 西北工业大学 High-resolution remote sensing image airplane detecting method based on high-level feature extraction of depth boltzmann machine

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20110142282A1 (en) * 2009-12-14 2011-06-16 Indian Institute Of Technology Bombay Visual object tracking with scale and orientation adaptation
CN102043958A (en) * 2010-11-26 2011-05-04 华中科技大学 High-definition remote sensing image multi-class target detection and identification method
CN104463248A (en) * 2014-12-09 2015-03-25 西北工业大学 High-resolution remote sensing image airplane detecting method based on high-level feature extraction of depth boltzmann machine

Cited By (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108734158A (en) * 2017-04-14 2018-11-02 成都唐源电气股份有限公司 A kind of real-time train number identification method and device
CN108734158B (en) * 2017-04-14 2020-05-19 成都唐源电气股份有限公司 Real-time train number identification method and device
CN107704509A (en) * 2017-08-31 2018-02-16 北京联合大学 A kind of method for reordering for combining stability region and deep learning
CN108805057A (en) * 2018-05-29 2018-11-13 北京师范大学 A kind of SAR image oil depot area detection method based on joint significance analysis
CN108805057B (en) * 2018-05-29 2020-11-17 北京师范大学 SAR image reservoir area detection method based on joint significance analysis
CN108985288A (en) * 2018-07-17 2018-12-11 电子科技大学 A kind of SAR image oil spilling detection method based on TGMSERs
CN108985288B (en) * 2018-07-17 2022-06-14 电子科技大学 TGMSERs-based SAR image oil spill detection method
CN109190457A (en) * 2018-07-19 2019-01-11 北京市遥感信息研究所 A kind of oil depot complex target rapid detection method based on large format remote sensing images
CN109190457B (en) * 2018-07-19 2021-12-03 北京市遥感信息研究所 Oil depot cluster target rapid detection method based on large-format remote sensing image
CN110674778A (en) * 2019-09-30 2020-01-10 安徽创世科技股份有限公司 High-resolution video image target detection method and device
CN113837270A (en) * 2021-09-18 2021-12-24 广东人工智能与先进计算研究院 Target identification method, device, equipment and storage medium
CN113837270B (en) * 2021-09-18 2022-08-30 广东人工智能与先进计算研究院 Target identification method, device, equipment and storage medium

Also Published As

Publication number Publication date
CN106557740B (en) 2019-05-03

Similar Documents

Publication Publication Date Title
CN115861135B (en) Image enhancement and recognition method applied to panoramic detection of box body
CN106557740B (en) The recognition methods of oil depot target in a kind of remote sensing images
CN110348319B (en) Face anti-counterfeiting method based on face depth information and edge image fusion
Wang et al. iVAT and aVAT: enhanced visual analysis for cluster tendency assessment
CN109684922B (en) Multi-model finished dish identification method based on convolutional neural network
CN108875600A (en) A kind of information of vehicles detection and tracking method, apparatus and computer storage medium based on YOLO
Yoon et al. LFIQ: Latent fingerprint image quality
CN107085708B (en) High-resolution remote sensing image change detection method based on multi-scale segmentation and fusion
CN107610114A (en) Optical satellite remote sensing image cloud snow mist detection method based on SVMs
CN107808138B (en) Communication signal identification method based on FasterR-CNN
CN108765465A (en) A kind of unsupervised SAR image change detection
CN108305260B (en) Method, device and equipment for detecting angular points in image
CN104408482A (en) Detecting method for high-resolution SAR (Synthetic Aperture Radar) image object
CN110415260B (en) Smoke image segmentation and identification method based on dictionary and BP neural network
CN110472081B (en) Shoe picture cross-domain retrieval method based on metric learning
CN109165658B (en) Strong negative sample underwater target detection method based on fast-RCNN
CN110728302A (en) Method for identifying color textile fabric tissue based on HSV (hue, saturation, value) and Lab (Lab) color spaces
CN106056139A (en) Forest fire smoke/fog detection method based on image segmentation
CN111274964B (en) Detection method for analyzing water surface pollutants based on visual saliency of unmanned aerial vehicle
CN108171119B (en) SAR image change detection method based on residual error network
CN108073940B (en) Method for detecting 3D target example object in unstructured environment
CN109726649A (en) Remote sensing image cloud detection method of optic, system and electronic equipment
CN110659374A (en) Method for searching images by images based on neural network extraction of vehicle characteristic values and attributes
CN113657196B (en) SAR image target detection method, SAR image target detection device, electronic equipment and storage medium
CN117037049B (en) Image content detection method and system based on YOLOv5 deep learning

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