CN107607203A - Conspicuousness band selection method based on structural similarity - Google Patents
Conspicuousness band selection method based on structural similarity Download PDFInfo
- Publication number
- CN107607203A CN107607203A CN201710803169.4A CN201710803169A CN107607203A CN 107607203 A CN107607203 A CN 107607203A CN 201710803169 A CN201710803169 A CN 201710803169A CN 107607203 A CN107607203 A CN 107607203A
- Authority
- CN
- China
- Prior art keywords
- band
- wave band
- wave
- structural similarity
- sub
- 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
Links
Landscapes
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The present invention relates to a kind of conspicuousness band selection method based on structural similarity, comprise the following steps:Step 1: using the target in hyperspectral remotely sensed image data Jing Guo denoising;Step 2: image all band is ranked up from high to low in the significant difference degree of each wave band by selected target region;Step 3: being screened two-by-two to Band Set after sequence based on structural similarity, the wave band that similarity is big and significance is low is removed;Step 4: selecting preceding some wave bands of wave band collection after screening, waveband selection is completed.The present invention is ranked up, and screen the larger wave band of similarity, the final preceding some wave bands chosen after screening by structural similarity to all wave bands accordingly using the significance apart from each wave band target area of notable transformation calculations.Such band selection method can obtain different waveband selections towards different target, preferably lift succeeding target Detection results.
Description
Technical field
The present invention relates to remote sensing fields, specifically make a kind of succeeding target Detection results preferably band selection method.
Background technology
Conspicuousness refers to based under human vision cognition system, target relative to background diversity factor, when target is more held
Easily notice that its brightness in notable figure is higher by human eye, it is also more notable.Significance has weighed the difference of target and background
Degree, it is clear that significance is higher, and target is easily detected.Vision significance conversion passes through range conversion function, by original
Figure is transformed to notable distance map, and each point pixel value is the significance of the point in artwork in notable figure.
If high spectrum image thinks input application, it is necessary first to rejects the redundancy of initial data, to believe in not weight losses
Reduce the dimension of data in the case of breath as far as possible, its main method is divided into waveband selection and the major class of feature extraction two, and wave band
Selection remains the physical meaning and the spectral charactersitics of targets of original wave band relative to feature extraction.Existing band selection method
Diversity factor between the information content of wave band itself and wave band is estimates, it is desirable to chooses that self-information amount is big, difference between each wave band
Spend big wave band.But existing method is not directed to specific objective information, the wave band for being more suitable for specific objective detection is chosen.
The content of the invention
The present invention is in order to overcome drawbacks described above present in prior art, there is provided a kind of significant difference degree waveband selection side
Method, calculate the significance of each wave band target area and all wave bands are ranked up accordingly, and phase is screened by structural similarity
Like larger wave band is spent, preceding some wave bands after screening are finally chosen.Such band selection method can towards different target,
Different waveband selections is obtained, preferably lifts succeeding target Detection results.
To solve the above problems, significant difference degree band selection method proposed by the present invention, comprises the following steps:
Step 1: using the target in hyperspectral remotely sensed image data Jing Guo denoising;
Step 2: selected target region is asked in the visual saliency of each wave band, according to each wave band visual saliency by up to
It is low that image all band is ranked up;
Step 3: being screened two-by-two to Band Set after sequence based on structural similarity, it is big and notable to remove similarity
Spend low wave band;
Step 4: selecting preceding some wave bands of wave band collection after screening, waveband selection is completed, obtains the bloom of wave band number reduction
Compose remote sensing image.
In above-mentioned technical proposal, the significant difference degree of foundation is asked for when the step 2 carries out band ordering, is employed aobvious
The range conversion of work property.The wave band screening that step 3 is carried out, estimating for two wave band similarity degrees is structural similarity.
In such scheme, significant difference degree band ordering specifically includes following sub-step in the step 2:
Sub-step S21, choose target region on image;
Sub-step S22, each wave band is regarded as different images, carries out view-based access control model conspicuousness conversion respectively;
Sub-step S23, each point significance in target region is averaged, obtain average significance;
Sub-step S24, according to the average significance of each wave band all wave bands of image are ranked up from high to low.
In such scheme, the screening of the wave band based on structural similarity of the step 3 specifically includes following sub-step:
Sub-step S31, shape similarity is asked for two-by-two to Band Set after sequence
Wherein μx、μyIt is the pixel average of x, y band image respectively;It is the variance of x, y band image, σxyIt is two images
Covariance.
Sub-step S32, screened according to threshold value, if structural similarity is more than given threshold, deletes in two wave bands and show
The smaller wave band of work degree, that is, sequence number is in wave band below
Beneficial effects of the present invention:
Conspicuousness band selection method proposed by the present invention based on structural similarity, for specific objective, to each wave band
The significant difference degree average of target area is sought, and all wave bands of image are ranked up according to average, the wave band so filtered out,
Target is all more prominent for background, and target detection effect is improved.Then redundancy is weeded out by shape similarity
Spend big wave band so that it is smaller to select band class information redundancy.
The present invention can finally obtain the image of wave band number reduction, and the image remains original image information well, and
More preferable effect can be obtained in target detection, i.e., the present invention processing after image can higher efficiency, in high quality carry out after
Continuous target detection.
Embodiment
Below in conjunction with specific embodiment, the present invention is described in further detail:
In the present embodiment, the conspicuousness band selection method proposed by the present invention based on structural similarity, including following step
Suddenly:
Step 1: using the target in hyperspectral remotely sensed image data Jing Guo denoising;
Step 2: selected target region is asked in the visual saliency of each wave band, according to each wave band visual saliency by up to
It is low that image all band is ranked up;
Step 3: being screened two-by-two to Band Set after sequence based on structural similarity, it is big and notable to remove similarity
Spend low wave band;
Step 4: selecting preceding some wave bands of wave band collection after screening, waveband selection is completed, obtains the bloom of wave band number reduction
Compose remote sensing image.
Significant difference degree band ordering specifically includes following sub-step in step 2:
Sub-step S21, the vehicle target for subsequently to be detected, optionally take one piece of area where target on image
Domain;
Sub-step S22, each wave band is regarded as different images, view-based access control model conspicuousness conversion carried out respectively, here using base
In the conspicuousness range conversion of minimum distance of obstacle
Wherein π () is from target point to figure
Some point being linked to be as edge background dot on path, f (π ()) are the pixel values of the point.Each pixel of each band image
A notable angle value of correspondence can be accessed.
Sub-step S23, each wave band is regarded as different images be handled as follows respectively:To vehicle target optional in image
Each pixel corresponds to significance and averaged in region, obtains average significance;
Sub-step S24, according to the average significance of each wave band all wave bands of image are ranked up from high to low.
In such scheme, the screening of the wave band based on structural similarity of the step 3 specifically includes following sub-step:
Sub-step S31, shape similarity is asked for two-by-two to Band Set after sequence
Wherein μx、μyIt is the pixel average of x, y band image respectively;It is the variance of x, y band image, σxyIt is two images
Covariance;
Sub-step S32, screened according to threshold value, selected threshold is 0.95 here, if structural similarity is more than setting threshold
Value 0.95, then the wave band that significance is smaller in two wave bands, that is, sequence number are deleted in wave band below.
Finally illustrate, the above embodiments are merely illustrative of the technical solutions of the present invention and it is unrestricted, although with reference to compared with
The present invention is described in detail good embodiment, it will be understood by those within the art that, can be to the skill of the present invention
Art scheme is modified or equivalent substitution, and without departing from the objective and scope of technical solution of the present invention, it all should cover at this
In the right of invention.
Claims (4)
1. a kind of conspicuousness band selection method based on structural similarity, it is characterised in that comprise the following steps:
Step 1: using the target in hyperspectral remotely sensed image data Jing Guo denoising;
Step 2: selected target region is asked in the visual saliency of each wave band, it is right from high to low according to each wave band visual saliency
Image all band is ranked up;
Step 3: being screened two-by-two to Band Set after sequence based on structural similarity, removal similarity is big and significance is low
Wave band;
Step 4: selecting preceding some wave bands of wave band collection after screening, waveband selection is completed, the EO-1 hyperion for obtaining wave band number reduction is distant
Feel image.
2. the conspicuousness band selection method according to claim 1 based on structural similarity, it is characterised in that the step
Wave band is ranked up preferably by target area visual saliency in rapid two, and step 3 screened by structural similarity it is similar
Spend big wave band.
3. the conspicuousness band selection method according to claim 1 based on structural similarity, it is characterised in that the step
Significant difference degree band ordering specifically includes following sub-step in rapid two:
Sub-step S21, choose target region on image;
Sub-step S22, each wave band is regarded as different images, carries out view-based access control model conspicuousness conversion respectively;
Sub-step S23, each point significance in target region is averaged, obtain average significance;
Sub-step S24, according to the average significance of each wave band all wave bands of image are ranked up from high to low.
4. the conspicuousness band selection method according to claim 1 based on structural similarity, it is characterised in that the step
Rapid three specifically include following sub-step:
Sub-step S31, shape similarity is asked for two-by-two to Band Set after sequence
Wherein μx、μyIt is the pixel average of x, y band image respectively;It is the variance of x, y band image, σxyIt is two images
Covariance;
Sub-step S32, screened according to threshold value, if structural similarity is more than given threshold, delete significance in two wave bands
Smaller wave band, that is, sequence number is in wave band below.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710803169.4A CN107607203B (en) | 2017-09-08 | 2017-09-08 | Conspicuousness band selection method based on structural similarity |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710803169.4A CN107607203B (en) | 2017-09-08 | 2017-09-08 | Conspicuousness band selection method based on structural similarity |
Publications (2)
Publication Number | Publication Date |
---|---|
CN107607203A true CN107607203A (en) | 2018-01-19 |
CN107607203B CN107607203B (en) | 2019-08-06 |
Family
ID=61063101
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710803169.4A Active CN107607203B (en) | 2017-09-08 | 2017-09-08 | Conspicuousness band selection method based on structural similarity |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107607203B (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116563571A (en) * | 2023-05-16 | 2023-08-08 | 北京师范大学 | Boltzmann entropy similarity-based hyperspectral image band selection method and system |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2007265248A (en) * | 2006-03-29 | 2007-10-11 | Toshiba Corp | Structured document management device, structured document sub management device, program and structured document managing method |
CN102096180A (en) * | 2009-12-11 | 2011-06-15 | 奥林巴斯株式会社 | Super-resolution microscope |
CN106204510A (en) * | 2016-07-08 | 2016-12-07 | 中北大学 | A kind of infrared polarization based on structural similarity constraint and intensity image fusion method |
CN106525238A (en) * | 2016-10-27 | 2017-03-22 | 中国科学院光电研究院 | Spaceborne multispectral imaging system design method based on super-resolution reconstruction |
-
2017
- 2017-09-08 CN CN201710803169.4A patent/CN107607203B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2007265248A (en) * | 2006-03-29 | 2007-10-11 | Toshiba Corp | Structured document management device, structured document sub management device, program and structured document managing method |
CN102096180A (en) * | 2009-12-11 | 2011-06-15 | 奥林巴斯株式会社 | Super-resolution microscope |
CN106204510A (en) * | 2016-07-08 | 2016-12-07 | 中北大学 | A kind of infrared polarization based on structural similarity constraint and intensity image fusion method |
CN106525238A (en) * | 2016-10-27 | 2017-03-22 | 中国科学院光电研究院 | Spaceborne multispectral imaging system design method based on super-resolution reconstruction |
Non-Patent Citations (1)
Title |
---|
李崇飞等: "基于结构相似度的视觉显著性检测方法", 《计算机工程与科学》 * |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116563571A (en) * | 2023-05-16 | 2023-08-08 | 北京师范大学 | Boltzmann entropy similarity-based hyperspectral image band selection method and system |
CN116563571B (en) * | 2023-05-16 | 2023-11-21 | 北京师范大学 | Boltzmann entropy similarity-based hyperspectral image band selection method and system |
Also Published As
Publication number | Publication date |
---|---|
CN107607203B (en) | 2019-08-06 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US10635929B2 (en) | Saliency-based method for extracting road target from night vision infrared image | |
CN107239759B (en) | High-spatial-resolution remote sensing image transfer learning method based on depth features | |
CN112184604B (en) | Color image enhancement method based on image fusion | |
CN106204510B (en) | A kind of infrared polarization and intensity image fusion method based on structural similarity constraint | |
CN103745449A (en) | Rapid and automatic mosaic technology of aerial video in search and tracking system | |
CN106846289A (en) | A kind of infrared light intensity and polarization image fusion method based on conspicuousness migration with details classification | |
CN101551853A (en) | Human ear detection method under complex static color background | |
CN106815817B (en) | A kind of improved high spectrum image denoising method | |
CN113298147B (en) | Image fusion method and device based on regional energy and intuitionistic fuzzy set | |
CN107609571A (en) | A kind of adaptive target tracking method based on LARK features | |
CN107527370A (en) | A kind of method for tracking target based on camshift | |
CN107784639A (en) | A kind of polygon filtering and noise reduction method of unmanned aerial vehicle remote sensing image improvement | |
CN104484670A (en) | Remote sensing image cloud detection method based on pseudo color and support vector machine | |
CN104361571B (en) | Infrared and low-light image fusion method based on marginal information and support degree transformation | |
CN104036461A (en) | Infrared complicated background inhibiting method based on combined filtering | |
CN107451986A (en) | Single infrared image enhancement method based on fusion technology | |
CN109635809B (en) | Super-pixel segmentation method for visual degradation image | |
CN109215006B (en) | Image fusion method based on total significant variation model and L2-norm optimization | |
Zhou et al. | New and Fast algorithms for Anomaly and Change Detection in Hyperspectral images | |
CN107607203B (en) | Conspicuousness band selection method based on structural similarity | |
CN109166088A (en) | Two waveband gray scale crater image fusion method based on non-down-sampled wavelet transformation | |
CN105528772B (en) | A kind of image interfusion method based on directiveness filtering | |
CN110111368B (en) | Human body posture recognition-based similar moving target detection and tracking method | |
CN106198402A (en) | A kind of basic rock recognition methods based on space flight Multi-spectral Remote Sensing Data | |
CN109949252A (en) | A kind of infrared image hot spot minimizing technology based on penalty coefficient fitting |
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 |